Outside Employment Opportunities, Employee Productivity, and Debt Disciplining Jayant R. Kalea, Harley E. Ryan, Jr.a, Lingling Wangb a

Department of Finance, J. Mack Robinson College of Business, Georgia State University, Atlanta, GA 30303, USA b

A.B. Freeman School of Business, Tulane University, New Orleans, LA 70118, USA

This version: September 27, 2011

Abstract We analyze how changes in labor market conditions influence the effect of a firm’s debt policy on its employee productivity and value. We first establish that debt in the capital structure increases the productivity of the firm’s employees and then show that this positive productivityleverage relation becomes weaker when outside employment opportunities for employees improve. We also find that the passage of NAFTA, an exogenous shock to employment opportunities in the manufacturing industry, strengthened the positive productivity-leverage relation for manufacturing firms. These effects are economically significant, and generally second only to the influence of asset intensity in magnitude. JEL classification: G30; G32; G38 Keywords: Debt Disciplining, Agency Theory, Outside Employment Opportunities, Employee Productivity

Contact info for Kale: +1 404 413 7345 and [email protected], for Ryan: +1 404 413 7337 and [email protected], and for Wang: +1 504 865 5044 and [email protected]. Kale acknowledges support from the H. Talmage Dobbs, Jr. Chair. We thank Vikas Agarwal, Rajesh Aggarwal, Thomas Bates, Bruce Carlin, Yongqiang Chu, Alex Edmans, Cheol Eun, Lorenzo Garlappi, Gerry Gay, Shingo Goto, Denis Gromb, Atul Gupta, Dirk Hackbarth, Kathleen Weiss Hanley, Iftekhar Hasan, Jean Helwege, Kose John, Marcin Kacperczyk, Simi Kedia, Omesh Kini, Kai Li, David Lesmond, Tanakorn Makaew, Steven Mann, Ron Masulis, Ernst Maug, Bill Megginson, George Morgan, Gordon Phillips, Josh Pierce, Eric Powers, Nagpurnanand Prabhala, Charu Raheja, Michael Rebello, Husayn Shahrur, Lemma Senbet, Jaideep Shenoy, Steve Smith, Paul Spindt, Sergey Tsyplakov, Chuck Trzcinka, Vahap Uysal, Anand Venkateswaran, Xiaoyun Yu, Donghang Zhang, and seminar participants at Georgia State University, Florida International University, Fordham University, Indian School of Business, University of Mannheim, University of South Carolina, University of South Florida, Corporate Finance Conference at Universitè Paris – Dauphine, 18th Conference on Financial Economics and Accounting at New York University, the Financial Management Association Meetings in Orlando, FL, and the Steve Smith Memorial Conference at the Federal Reserve Bank of Atlanta for comments. We thank the Bureau of Labor Statistics and John McLaren and Shushanik Hakobyan for providing some of the data used in our analysis. We alone are responsible for remaining errors.

Outside Employment Opportunities, Employee Productivity, and Debt Disciplining

I.

Introduction By some estimates, the North American Free Trade Agreement (NAFTA) resulted in a decrease of more than 879,000 actual and potential jobs between 1994 and 2002; mostly in the manufacturing sector (Scott, 2003). Although the net merits of NAFTA are still under debate, it is clear NAFTA affected the actual and perceived employment opportunities in certain industries (McLaren and Hakobyan, 2010). In this paper, we seek to answer a basic empirical question: How do changes in (or shocks to) labor markets impact the efficacy of a firm’s financial policies? Specifically, we analyze the effects of changes in outside employment opportunities (including those arising from the passage of NAFTA) on the influence of a firm’s debt policy on its productivity and, consequently, its value. Our research is motivated by the insights provided by two strands of literature. One strand consists of theoretical and empirical studies that propose and document relations between capital structure and conditions in the labor market.1 The second strand emanates from the seminal papers by Grossman and Hart (1982) and Jensen (1986), which demonstrate that risky debt serves as a costly disciplining mechanism to reduce the agency costs that arise from the separation of ownership and control. Our paper spans these two strands by examining how changes in employment opportunities in the labor market affect the disciplining role of debt. The rationale underlying the disciplining role of debt is that the presence of risky debt in the firm’s capital structure introduces the possibility that the firm goes bankrupt. The manager of the firm receives benefits/payoffs that relate positively to firm value in non-bankrupt states but 1

For the effect of labor markets on capital structure, see Bronars and Deere (1991), Perotti and Spier (1993), Dasgupta and Sengupta (1993), Hanka (1998), Chen, Kacperczyk, and Ortiz-Molina (2010, 2011), Matsa (2010), Berk, Stanton, and Zechner (2010), and Agarwal and Matsa (2011).

2 bears significant personal costs in the event the firm goes bankrupt. In this setting, debt provides the manager an incentive to increase his effort, which reduces the likelihood of bankruptcy and results in higher firm value. We interpret managerial effort broadly and an increase in manager’s effort can include working harder, motivating junior employees, consuming fewer perquisites, optimizing span of control, and making more efficient use of capital and labor inputs. For example, in our setting, higher managerial effort will include cases where the manager lays of less productive workers or effects production cost savings. We hypothesize that this disciplining role of debt becomes weaker if agents have alternative employment opportunities. The agent will optimally choose to accept a job in another firm if the cost of changing employers is lower than the disutility of extra effort and expected personal bankruptcy costs associated with a higher debt level in the current firm.2 Thus, the disciplinary role of debt will be weaker when the agent has more outside employment opportunities. We examine a sample of over 103,000 firm-years from 1976 to 2007 and find strong support for the disciplining role of debt and our hypothesis that better outside employment opportunities reduce the efficacy of debt as a disciplining mechanism. To our knowledge, our study is the first large-sample examination of debt as a disciplining mechanism and also the first to incorporate explicitly outside employment opportunities into the analysis. Grossman and Hart (1982) and Jensen (1986) demonstrate that risky debt disciplines top managers. In our study, we relate debt to the productivity of all the employees in the firm. There are several reasons for our choice. First, the actions of top management are generally not observable, which makes it difficult to measure the productivity of top managers in isolation.

2

The presence of debt may also reduce the ability of the manager to consume perks and/or invest in “pet” projects, and increase the disutility from additional monitoring. See, for instance, Stulz (1990).

3 Second, total employee productivity is a measurable outcome that is highly correlated with top management effort because top managers direct much of their effort towards making other employees more productive. Empirical studies indicate that top management spends 50% to 64% of its time in verbal contact with subordinates (see Kurke and Aldrich (1983) for a summary of the evidence). We argue that debt provides top management with the incentive to exert more effort (broadly defined) to closely monitor middle managers, who then expend greater effort to monitor lower-level employees, and so on down the corporate hierarchy. Thus, total employee productivity should be highly correlated with top management effort. As Brealey, Myers, and Allen (2006, p. 10) note: [j]ust as shareholders need to encourage managers to work for the shareholders’ interests, so senior management needs to think about how to motivate everyone else in the company. Third, since CEOs are likely to lose their jobs when their firms experience financial distress (Gilson, 1989, 1990; Gilson and Vetsuypens, 1993), they have a strong incentive to take preemptive actions such as layoffs, which result in job losses for lower-level employees. Consistent with this view, Sharpe (1994) shows that the cyclicality of the labor force in manufacturing firms is positively related to financial leverage. So, higher debt should have a direct effect on the productivity of all firm employees since the possibility of financial distress creates job-loss concerns for all the employees of a firm. To test our hypothesis that better outside employment opportunities reduce the efficacy of debt as a disciplining mechanism, we examine the influence of changes in outside employment opportunities on the relation between employee productivity and leverage. We use two measures of employee productivity: (i) output per employee and (ii) output per employee hour; and three measures of industry-level outside employment opportunities: (i) the voluntary quit rate (the

4 proportion of employees in an industry that voluntarily quit their jobs) (ii) the hire rate (the number of new hires in an industry divided by total employees in the industry) and (iii) the rate of growth in unemployment in the firm’s industry. In addition to these industry-specific variables, we also use the passage of the NAFTA in 1994 as an exogenous shock to employment opportunities, primarily in the manufacturing sector.3 We examine a panel data set of 103,394 firm-years encompassing more than 13,800 firms from 1976-2007. Consistent with the debt disciplining hypothesis, we find a significantly positive relation between employee productivity and financial leverage. We then investigate how employment opportunities affect the strength of the positive productivity-leverage relation. The findings for all of the productivity and outside employment opportunity proxies tell a consistent story: When outside employment opportunities increase (decrease), the relation between employee productivity and debt is significantly weaker (stronger). When an exogenous event, NAFTA, reduced the employment opportunities for employees, we find that the relation between productivity and leverage became stronger. That is, the passage of NAFTA strengthened the disciplining role of debt. Further, we show that the effect of NAFTA occurs mainly in manufacturing industries, which were more susceptible to job losses (Scott, 2001). Within the sample of manufacturing industries, we find that NAFTA had greater effect on industries where high tariff rates made them more vulnerable to competition (see McLaren and Hakobyan, 2010). Our findings are robust to a variety of econometric approaches, specifications, and tests to rule out additional alternative explanations for our results. Further, the impact of leverage on productivity is economically significant. We show that only asset intensity has a consistently greater impact than leverage on employee productivity and firm value. Taken together, our 3

Scott (2001, 2003) and Hottenrott and Blank (1998) present evidence that NAFTA resulted in significant job losses and affected job opportunities primarily in the manufacturing sector.

5 empirical findings support the hypothesis that debt serves as a costly disciplining mechanism and that greater outside employment opportunities reduce the effectiveness of this disciplining mechanism. We also consider whether our results derive from a composition effect instead of an incentive effect. It is possible that unobserved omitted variables simultaneously influence financial conditions, employment opportunities, and product market conditions. If highly levered firms are more sensitive to these forces, absent any agency problems, managers would optimally alter the composition of inputs to maximize productivity without any incentive from leverage. To address this possibility, we separately examine firms with high- and low-leverage and find that our relations hold in both subsamples. Another possibility is that leverage provides top managers the incentives to alter the composition of inputs but there is no incentive effect for the other employees to increase effort. For instance, following the NAFTA shock, firm productivity may increase simply because managers of highly levered firms lay off the least productive employees. Union collective bargaining contracts frequently call for layoffs by reverse seniority and, as a result, layoffs of less productive workers are less likely to occur when the labor force is unionized. Therefore, we divide the firms in our sample by their industries’ union coverage and verify that our findings hold in highly unionized industries as well as industries with low union coverage. Although these tests do not rule out the possibility that changes in the composition of inputs influence our findings, the aggregate evidence suggests that debt also provides incentives to all employees. Our finding of a positive relation between employee productivity and financial leverage adds to the empirical evidence on the disciplining role of debt. Most existing studies in this area focus on firms that are targets of leveraged buyouts or engage in highly levered recapitalizations.

6 Although these studies offer support for the disciplining role of debt, the sample sizes in these studies are rather small since firms seldom conduct such highly-leveraged recapitalizations.4 More broadly, our finding that greater outside employment opportunities weaken the relation between employee productivity and financial leverage adds to our understanding of how financial policies interact with nonfinancial markets. In support of the debt-disciplining hypothesis, Phillips (1995) finds that inefficient output in the product markets declines following leveraged recapitalizations. Hanka (1998) and Matsa (2010) show that financial leverage can be used as a strategic tool to negotiate with labor. Agarwal and Matsa (2011) demonstrate the effect of unemployment risk to employees on a firm’s capital structure, and Chen, Kaperczyk, and Ortiz-Molina (2010, 20011) find that labor unions affect a firm’s cost of capital. Our findings complement this literature by showing that labor market conditions influence the efficacy of debt disciplining. Our study also contributes to the broader literature that examines the interactions between corporate capital structure and financial and non-financial stakeholders.5 By documenting that more outside employment opportunities weaken the ability of debt to discipline the agents of the shareholders, we complements research on how employment opportunities affect firms’ attempts to mitigate agency problems. For instance, Cappelli and Chauvin (1991) provide evidence that fewer alternative employment opportunities result in fewer incidences of employee shirking; Parrino (1997) provides evidence that outside employment opportunities influence the decision to fire and hire CEOs; and consistent with Oyer (2004),

4

Kaplan (1989), Lehn and Poulsen (1989), Marais, Schipper, and Smith (1989), Lehn, Netter, and Poulsen (1990), Muscarella and Vetsuypens (1990), Smith (1990), Opler (1992), Opler and Titman (1993) and Denis and Denis (1993) are studies that investigate the disciplining role of debt. 5 See, for instance, Titman (1984), Brander and Lewis (1986), Maksimovic (1988, 1990), Chevalier (1995), Kovenock and Phillips (1995, 1997), Campello (2006), and Kale and Shahrur (2007).

7 Rajgopal, Shevlin, and Zamora (2006) document that outside employment opportunities influence the structure of CEO incentive compensation. The rest of the article is organized as follows. The next section describes our data sources and the variables we use in the analysis. Section III discusses our empirical method. Section IV presents our primary empirical findings, and Section V presents results from robustness tests, and the last section concludes. II.

Data, variable construction and summary statistics A. Data Our initial sample includes all firms in the Compustat Industrial Annual database from 1976 to 2007. We exclude ADRs, financial (SIC codes 6000 to 6999) and utility firms (SIC codes 4900 to 4999). We also exclude firms with sales or asset growth rates of more than 200% a year since such high growth usually signals merger or acquisition activities that could affect employee productivity and financial leverage (see Campello, 2006).6 Because financially distressed firms tend to perform poorly (e.g. Opler and Titman, 1993), we also exclude firms with debt ratio in the top debt decile of our sample (debt ratio above 0.671).7 Strebulaev and Yang (2006) find that zero-debt firms are smaller, more profitable, and have higher dividend payout ratios and cash balances than industry-matched firms with debt. Firms with high growth opportunities may also optimally carry no debt to avoid the Myers (1977) debt overhang problem (Gardner and Trzcinka, 1992). Thus, we exclude zero-debt firms from our main tests, but verify that our results hold if we include zero-debt firms. Our results also hold if we exclude only utility and financial firms and do not impose any additional restrictions on the Compustat sample. 6

As an alternative, we repeat all tests on a sample in which we exclude firm-years in which firms reported M&A sales impact in the income statement, and find similar results. 7 Our results are robust if we exclude firms in the top debt quartile. As another check, we include firms in the top leverage decile and estimate a quadratic leverage model to show the appropriateness of the sample restriction.

8 We use data from the Job Opening and Labor Turnover Survey (JOLTS) and the Current Population Survey (CPS) provided by the Bureau of Labor Statistics (BLS) to compute measures of outside employment opportunities (quit rate, hire rate, and unemployment growth rate). The JOLTS data provide information on the quit rate and hire rate from 2000 and covers all nonagricultural industries.8 The CPS survey provides data on the unemployment rate from 1976 and covers all industries. In our tests, we adjust variables by subtracting their respective industry medians or controlling for industry fixed effects. To be consistent with the industry definitions from the JOLTS data, we report results based on 2-digit SIC codes and require a minimum of 5 firms in the firm’s 2-digit SIC code. Our results are robust if we base industry medians or fixed effects on 3-digit SIC codes. We obtain accounting data from the Compustat database and require information on the number of employees, total assets, debt, sales and profitability.9 We use the Consumer Price Index (CPI-U) compiled by the BLS to adjust dollar values to 2003 dollar levels. To control for the influence of extreme values, we winsorize firm-level variables by setting values that exceed the 99th percentile or fall below the first percentile to the 99% and 1% values, respectively. Our final sample consists of 103,394 firm-years for 13,829 firms; the number of firms ranges from 2,319 to 4,222 per year over the sample period.

8

The BLS also provides quit rates and hire rates from 1970 to 1981. Since these data cover only the manufacturing industry and are not directly comparable with the JOLTS data, we use the quit rates and hire rates from JOLTS data in the main tests. We verify that our results hold if we use the the BLS quit rates and hire rates from 1970-1981. 9 We obtain similar results for a sample that requires a minimum of ten firms in an industry. Since historical SIC codes are available in Compustat only from 1987, we use the 1987 historical SIC code for years prior to 1987.

9 B. Variable Construction B.1 Outside Employment Opportunities We use three measures of changes in outside employment opportunities: (1) industry Unemployment Growth Rate (2) Quit Rate by industry and (3) Hire Rate by industry. In addition, we investigate the effect of NAFTA, an exogenous shock to employment opportunities, on the leverage-productivity relation. The Unemployment Growth Rate, which measures the marginal change in industry employment opportunities, is the percentage change in the industry unemployment rate.10 We use the change in unemployment and not the level of unemployment since, in equilibrium, debt and effort levels already reflect unemployment levels. The BLS defines the Quit Rate as the number of employees who leave their jobs voluntarily during the survey period (except retirements or transfers within the same firm) divided by the number of employees in the industry. The Hire Rate is the number of new hires in the industry during the survey period divided by the number of employees in the industry. A higher Quit Rate generally implies better outside employment opportunities for employees in that industry. However, the Quit Rate could be higher if employees voluntarily quit their jobs to pursue opportunities in other industries. The Hire Rate does not suffer from this possibility. The Quit Rate and Hire Rate are available from 2000-2007 for 2-digit SIC code industry classifications. We also investigate the effect of an exogenous shock to employment opportunities, the passage of NAFTA in 1994, on the leverage-productivity relation. Scott (2003) estimates that from 1994 through 2002, NAFTA increased the trade deficit with Canada and Mexico and

10

The CPS Survey in BLS provides unemployment rates for 2-digit SIC codes before 2002 and 3-digit NAICS codes after 2002. We assign the unemployment rate to a firm based on its 2-digit SIC industry code before 2002 and 3-digit NAICS industry code after 2002.

10 “caused the displacement of production that supported 879,280 jobs.” To examine the effect of NAFTA, we create a dummy variable, NAFTA, which equals zero if the firm-year is before 1994 and one if the firm-year is 1994 or later. Research (Scott, 2001, 2003; Hottenrott and Blank, 1998) shows that the passage of NAFTA affected job opportunities primarily in the manufacturing sector. McLaren and Hakobyan (2010) use Mexico tariff data and verify that the influence of NAFTA varied significantly by industry. Thus, we also conduct tests on subsamples of firms in manufacturing and non-manufacturing sectors, as well as the most and least affected industries by tariff changes. B.2 Employee Productivity and Imputed Firm Value We use two primary measures of firm-level employee productivity: Output per Employee and Output per Employee Hour. The variable, Output per Employee, is the ratio of firm output to the number of employees in the firm (Compustat data item 29). We follow Schoar (2002) and Brynjolfsson and Hitt (2003) and measure the firm’s output as sales (data item 12) plus changes in inventories (data items 77 and 78). Output per Employee Hour is the ratio of output to the total number of employee hours worked. To estimate the total number of employee hours worked, we multiply the number of employees in the firm by the average number of annual labor hours per employee in the firm’s industry, which is calculated from the industry employment data published by the BLS.11 The advantage of Output per Employee is that we can uniquely identify it for each firm. However, this measure could be distorted by firm strategies. For instance, firms could respond to changes in product demand, which could correlate highly with changes in employment opportunities, by paying employees to work overtime. Thus, an increase in Output per Employee would be a misleading since employees work more hours. Output per Employee 11

The BLS reports data across sectors at different industry definitions (e.g., number of SIC or NAICS digits). For each firm, we use the finest industry definition available in the data.

11 Hour avoids this possibility, but it is not available for all firms in our sample. Our results are also robust to using a value-added productivity measure, EBITDA per Employee. To examine the economic significance of the influence of leverage on productivity, we construct a measure of imputed firm value based on employee productivity. We compute Imputed Firm Value as follows. For each firm-year, we estimate the firm value as the sum of the market value of equity (data item 25*data item 199) and the book value of debt (data item 9 plus data item 34). We then divide firm value by Output per Employee (Employee Hour) to compute the ratio of firm value per unit of productivity. We use the 2-digit SIC code industry median ratio of value to productivity as a value multiplier to estimate the value that comes from industry productivity. For each firm-year, we estimate Imputed Firm Value by multiplying Output per Employee (Employee Hour) by the median industry ratio of value to productivity.12 Imputed Firm Value provides an estimate of value increases induced by changes in productivity; whereas total firm value can be influenced by other factors such as debt tax shields or changes in growth opportunities. B.3 Leverage We measure Leverage as the book value of long-term debt plus debt in current liabilities (data item 9 + data item 34) divided by book value of debt plus the market value of equity (data item 9 + data item 34 + data item 25*data item 199). For robustness, we repeat all our tests with Leverage computed with the book value of equity and obtain similar results.

12

Suppose that a firm’s output per employee is $239,000 and the industry median value multiplier (value per unit of productivity) is 770. The Imputed Firm Value - Employee is $239,000 X 770 or $184.03 million.

12 B.4 Control Variables We control for a number of other factors that could influence productivity. We include the number of employees (date item 29) as a proxy for labor input. Following Hanka (1998), we include the variable Asset Intensity, total assets (data item 6) divided by the number of employees, as a proxy for capital input per employee. Since productivity could increase along a learning curve as firms mature, we include the variable Firm Age, which is the number of years a firm has been in the Compustat annual database. Since the Compustat database contains data on firms dating to 1950, the maximum value of Firm Age is 57 years. We include the variable Operating Leverage, the ratio of gross property, plant & equipment (data item 7) to total assets, to control for the possibility that operating leverage influences productivity.13 To control for the effects of product market competition (Jensen, 1986; Maksimovic, 1988; Philips, 1995; Mackay and Philips, 2005), we use the firm’s industry fitted Herfindahl-Hirschman Index (HHI), which is based on both public and private firms (Hoberg and Phillips, 2010). B.5 Descriptive Statistics We present summary statistics for the sample in Panel A of Table 1. Summary statistics for our primary test variables – leverage, employee productivity measures, and outside employment opportunities – are in the top portion of Panel A. Since data from the JOLTS are available only from year 2000, the sample size declines from 103,394 firm-years for Leverage and Output per Employee to 25,250 for Quit Rate and Hire Rate. The mean Leverage is 0.238 and the median is 0.200. The mean and median Output per Employee are $239,154 and $163,711, respectively. The mean Output per Employee Hour is $127.234 and the median is $85.709. The mean (median) values for the outside employment opportunity variables, Quit Rate

13

We obtain similar results if we use net property, plant & equipment divided by total sales.

13 and Hire Rate, are 1.816% (1.508%) and 3.527% (2.967%), respectively. The mean and median values for Unemployment Growth Rate are 1.511% and -3.488%, respectively.14 We present descriptive statistics for control variables in the bottom portion of Panel A. The minimum number of employees in a firm is 3 and the maximum number of employees in a firm is 116,192. To explore the possibility that our results are driven by firms at either extreme, we conduct our analysis on a sample without firms in the bottom (fewer than 60 employees) and top (more than 15,991 employees) deciles for the number of employees and find similar results. In Panel B of Table 1, we report correlations between the leverage, productivity, and outside employment opportunity variables. Consistent with the disciplining role of debt, there is a significantly positive correlation between Leverage and all employment productivity measures. The measures of outside employment opportunities, Quit Rate and Hire Rate are positively correlated with each other. Since Unemployment Growth Rate measures a lack of outside employment opportunities, it correlates negatively with Quit Rate and Hire Rate. B.6 Employee Productivity by Industry To examine the variation in employee productivity across industries, we present in Table 2 the median values for leverage, productivity, and outside employment opportunities variables in each 2-digit SIC code industry. The table presents industries according to decreasing median values of Output per Employee. Productivity varies considerably across industries. For both Output per Employee and Output per Employee Hour, the “Petroleum and Coal Products” industry has the highest median and “Social Services” has the lowest. The median Leverage is highest for “Automotive Repair, Services, and Parking” (0.45) and lowest for “Business

14

We do not winsorize industry-level variables but we find in unreported tests that all results hold if we winsorize the three outside employment proxies at their respective 1% and 99% values.

14 Services” (0.08). The highest median outside employment opportunities are in “Automotive Repair, Services, and Parking” and “Hotels, Rooming Houses, Camps, and Other Lodging” for Quit Rate (4.80), “Motion Pictures” for Hire Rate (7.29), and “Tobacco Manufacturers” for Unemployment Growth Rate (4.60). The lowest medians are in “Educational Services” for Quit Rate (1.16), “Miscellaneous Repair Services” (2.20) for Hire Rate, and “Primary Metal Industries” and “Paper and Allied Products” (-8.89) for Unemployment Growth Rate. Given the considerable variation in employee productivity across industries, we use two methods to control for industry effects. In the first method, we adjust all firm-level continuous variables, except Leverage, by subtracting the 2-digit SIC code industry median from the variable. We do not adjust Leverage by its industry median because, ceteris paribus, it is not clear if the likelihood of financial distress depends on the level of debt in the firm or the industry-adjusted level. For completeness, however, we repeat our median-adjusted regressions using industry median-adjusted leverage and find similar results to those reported in the tables. In the second method, we include industry dummy variables in our specification to control for industry fixed effects, which adjusts all variables from the respective industry mean. III.

Empirical model specification Our objective is to examine how the disciplining role of debt is affected by the changes in outside employment opportunities. To establish a benchmark, we first establish the relation between employee productivity and leverage without considering outside employment opportunities and then modify the benchmark model to study the effects of outside employment opportunities on the leverage-productivity relation. The regression specification for the benchmark productivity-leverage relation is as follows:

15 E m p lo y e e P r o d u c tiv ity it   0   1 L e v e r a g e it

1





 k C o n tr o l V a r ia b le s   it

(1)

kK

The control variables include natural log values of Asset intensity, Employees, and Firm Age. We also control for Operating Leverage, Herfindahl Index, and year fixed effects. If the debtdisciplining hypothesis is correct, the coefficient on Leverage, β1, should be positive. We then examine if and how changes in outside employment opportunities affect the relation between debt and employee productivity. To this end, we include interaction variables between different measures of outside employment opportunities and Leverage. Specifically, we estimate the following augmented version of equation (1): E m p lo y e e P r o d u c tiv ity it   0   1 L e v e r a g e it

1



 2 L e v e r a g e it * O u ts id e E m p lo y m e n t O p p o r tu n itie s 1

  3 O u ts id e E m p lo y m e n t O p p o r tu n itie s 



(2)

 k C o n tr o l V a r ia b le s   it

kK

In the above specification, outside employment opportunities are the Quit Rate, the Hire Rate, or the Unemployment Growth Rate. Our hypothesis states that if the outside employment opportunities increase (decrease), then the relation between productivity and leverage will become weaker (stronger). Therefore, we expect that the coefficient 2 on the interaction of Leverage and the outside employment measure will be negative for proxies where higher values represent an increase in outside employment opportunities (e.g., quit rate and hire rate) and positive for proxies where higher values represent a decrease in outside employment opportunities (e.g., unemployment growth rate). To address the possibility that endogeneity or unobservable omitted variables cause leverage, employee productivity, and outside employment opportunities to move together and result in a spurious relation, we (i) use lagged leverage in the regressions; (ii) examine the impact

16 of an exogenous shock, NAFTA, on the productivity-leverage relation; (iii) conduct a residual analysis in which we substitute the residual from a regression of outside employment opportunity proxies on leverage for the outside employment opportunity proxies. IV.

Effects of outside employment opportunities on the leverage-productivity relation In this section, we present the results from our estimates of the benchmark LeverageProductivity relation (equation (1)), and how this relation is affected by outside employment opportunities (equation (2)). We present results first for industry median adjusted variables and then for industry fixed effects. We then present results for the effect of the NAFTA shock on the Productivity-Leverage relation, and subsample analyses based on NAFTA-induced tariff changes, high and low leverage, and degree of union coverage. At the end of the section, we use the Imputed Firm Value as the dependent variable to present evidence on the economic significance of our findings. A. Effects of cross-sectional measures of outside employment opportunities In Table 3, we present evidence on the benchmark Leverage-Productivity relation and the effect of the Quit Rate, Hire Rate, and Unemployment Growth Rate, on this relation. Higher values for the Quit Rate and Hire Rate imply better employment opportunities in the firm’s industry, and a higher Unemployment Growth Rate implies worse employment opportunities in the firm’s industry. All firm-level variables, except for Leverage, are adjusted for the respective industry medians. We present robust standard errors based on firm clustering in parentheses. The first four columns present analyses based on Output per Employee. The coefficient on Leverage in column one is positive (0.143) and statistically significant at the 1% level, which is consistent with the hypothesized disciplining role of debt. In column two, the coefficient on

17 the interaction term between Leverage and the Unemployment Growth Rate is significantly positive (coefficient = 0.176, standard error = 0.032) which supports our hypothesis that the disciplining effect of Leverage becomes more effective as employment opportunities worsen. Using the Quit Rate and Hire Rate as indications of better employment opportunities confirms this result. The coefficient on the interaction term of Leverage and the Quit Rate is negative (coefficient = -0.139), as is the coefficient on the interaction between Leverage and the Hire Rate (-0.092); both relations are significant at the 1% level. The findings presented in the last four columns based on Output per Employee Hour tell a similar story. The coefficient on Leverage in column five of the table is significantly positive (coefficient = 0.169, standard error = 0.021). The coefficient on the interaction term with Unemployment Growth Rate is significantly positive, and the coefficients on the interactions with Quit Rate and Hire Rate are significantly negative. Table 4 presents the findings from the industry fixed-effects approach. These findings are similar to those reported for the industry median-adjusted approach. The significantly positive coefficients on Leverage in columns one (for Output per Employee) and five (for Output per Employee Hour) support the hypothesized disciplining role of debt. The coefficients on the interaction term between Leverage and Unemployment Growth Rate in columns 2 and 6 are positive, and the coefficients on the interaction terms between Leverage and the Quit Rate and Leverage and the Hire Rate (columns 3, 4, 7, and 8) are all negative. The coefficients are all significant at the 1% level, except for the interaction between Leverage and the Quit Rate for Output per Employee Hour, which is significant at the 10% level. Since the median-adjusted approach and the industry fixed-effects approach yield similar results, we present only results for the median-adjusted approach in the remainder of the paper.

18 B. The effect of an external shock to outside employment opportunities The passage of NAFTA in 1994 was an external shock to labor markets in many industries (McLaren and Hakobyan, 2010), and thus allows us to test our prediction that the efficacy of debt as a disciplinary mechanism is affected by changes in outside employment opportunities available to employees, particularly those in the manufacturing sector (Scott, 2003, 2001; Hottenrott and Blank, 1998).

Since NAFTA reduced employment opportunities, we

predict that the disciplining role of debt is stronger in the period after its passage in 1994. We analyze the entire sample as well as subsamples of manufacturing (SIC codes 2000 – 3999) and non-manufacturing firms. Following McLaren and Hakobyan (2010), we also use changes in Mexico tariff rates to isolate the most- and least-affected industries. We present the findings from our analysis of NAFTA as an exogenous shock to employment opportunities in Table 5. The first three columns present results based on Output per Employee and the second three columns present results based on Output per Employee Hour. In each set of three columns, we present results based on the entire sample, the subset of firms in the manufacturing sector, and the subset of firms in the non-manufacturing sector. The coefficient on Leverage captures the effect of Leverage on productivity when NAFTA is zero, that is, in the period prior to the passage of the act in 1994. The coefficient on Leverage is significantly positive at the 1% level in all columns, which indicates that the LeverageProductivity relation is positive prior to NAFTA.15 As predicted, the coefficients on the interaction term between Leverage and NAFTA are significantly positive at the 1% level for the entire sample and the manufacturing sector. These findings are similar for both productivity measures and offer strong support for our hypothesis that a worsening of job opportunities 15

An F- test on the joint significance of Leverage and the interaction term with NAFTA indicates that total Productivity-Leverage relation is significantly positive also after the passage of NAFTA.

19 increases the disciplining power of debt. Consistent with evidence that suggests NAFTA reduced employment opportunities primarily in the manufacturing sector, we do not find a significant interaction effect for non-manufacturing firms. McLaren and Hakobyan (2010) document that the impact of NAFTA on labor markets varied significantly by industry, but find only modest effect on geographic locations that were expected to be vulnerable to NAFTA. They note that “[it] is better to be a software worker in a textile-and-apparel town than a textile-and-apparel worker in a software town.” Thus, further segregating the sample according to the most and least vulnerable industries provides an additional means of testing our hypothesis. We follow McLaren and Hakobyan (2010) to calculate imports-weighted average Mexico tariff rates for every industry. The data on Mexico tariff rates and imports are originally from the U.S. Census Bureau and mapped to 1987 SIC codes for manufacturing industries (see Schott (2010) for details). As in McLaren and Hakobyan, we first calculate the average industry Mexico tariff rates for 1990 and 2000, and estimate the changes in these rates following the passage of NAFTA by subtracting the rates in 1990 from those in 2000. We use these changes in tariff rates to classify manufacturing firms into those most and least affected by NAFTA. For our sample, the mean (median) change in Mexico tariff rates is -1.93% (-1.60%). Over 86% of the manufacturing firms are in industries that faced reduced tariffs following NAFTA. Table 6 presents the results of our analysis based on subsets of the manufacturing sector sample segregated by changes in tariff rates. The results further confirm our hypothesis that the disciplining effect of leverage becomes stronger as employment opportunities worsen. For the least vulnerable industries (the 30% with the least reduction in tariff), we find no influence of NAFTA on the leverage-productivity relation. For the most vulnerable industries (top 30%), we

20 find a positive but insignificant relation between Leverage and productivity before NAFTA, and a positive) Leverage-productivity relation after NAFTA, significant at the 1% level. The effect for moderately vulnerable industries (middle 40% by tariff reduction) is also positive and significant. We also divide the sample into firms with a reduction in tariff rates and either no change or increase in tariff rates. Providing additional support for our proposition, we document that the coefficient on the interaction between NAFTA and Leverage is positive and significant at the 1% level when tariffs decrease, but is not different from zero when tariffs remain stable or increase. McLaren and Hakobyan (2010) identify only 12 non-manufacturing industries that have tradable goods with Mexico. The data indicate that the impact of NAFTA on employment opportunities is substantially weaker for these non-manufacturing industries.

For these 12

industries, the average reduction in tariff from 1990 to 2000 is 0.60% compared about a 2% reduction, on average, for manufacturing industries. Six of the non-manufacturing industries have no tariff reduction. Thus, since the changes in tariffs are minimal, we do not expect a strong NAFTA effect. We use the data from McLaren and Hakobyan to estimate the impact of NAFTA on the leverage-productivity relation for non-manufacturing firms and present the results in Table 7. As expected, the influence is weaker, but the sign on the interaction between NAFTA and Leverage is positive when there is a tariff reduction, significant at the 10% level when we use Output per Employee Hour to measure productivity. We find no influence when there is no reduction in the tariff rate. C. Alternative interpretations and additional analysis We recognize the possibility that the financial condition of the firm, employment opportunities, and productivity interact as endogenous responses to a change in some unobserved

21 economic variable. For instance, suppose that macroeconomic changes simultaneously affect product market and labor market conditions and that all firms react optimally to change the composition of their inputs. Firms with high financial leverage might respond more aggressively to the macroeconomic changes and optimally change the composition of their employees to increase labor productivity. Such a change need not represent debt disciplining.

It is also

possible that top managers respond to debt disciplining by firing the least productive workers, leaving a more productive workforce. Although this explanation is consistent with debt disciplining at the top levels, it raises the possibility that debt does not provide incentives for all employees. As an exogenous shock to employment opportunities, NAFTA provides an opportunity to examine these possibilities. To examine the possibility that our results merely reflect interactions between product markets, labor markets, and the financial condition of the firm, we divide the manufacturing sample into high-leverage firms (above the sample median) and low-leverage firms (below the sample median). If the alternative hypothesis is correct, we would expect our results to concentrate in the high-leverage firms. To examine the possibility that upper management simply changes the composition of employees by firing the least productive employees, we divide the manufacturing sample into industries with more (above the median) and less (below the median) union coverage in 1994. We obtain industry union coverage from the Current Population Survey (CPS) union database (www.unionstats.com) compiled by Hirsch and Macpherson (2003) and use their matching table to match the CPS industry codes to either the 4- or 3-digit SIC code depending on availability. Although union and non-union firms alike are subject to the threat of layoffs, which provides an incentive for all employees to exert more effort, union contracts

22 typically specify that layoffs occur by reverse seniority. Thus, if our results simply reflect a composition effect, we would expect the results to concentrate in the least unionized firms. Table 8 presents our analysis of the influence on NAFTA on the leverage-productivity relation in manufacturing firms, segregated by level of (i) financial leverage and (ii) union coverage For both productivity measures, the coefficient on the interaction between Leverage and NAFTA is positive and significant at the 1% level for both high- and low-leverage firms. Thus, it is unlikely that our results merely represent the response of highly-levered firms to unobserved variables. When we divide the sample by union coverage, we find that the interaction between Leverage and NAFTA is positive and significant at the 5% or 1% level for firms with high union coverage and low union coverage. Thus, it is unlikely that our results reflect only a composition change where top managers fire the least productive employees. We believe that some change in the composition of employees is likely and consistent with debt disciplining. Indeed, increased debt discipline provides top management with the incentive to modify the workforce in response to changing conditions, and some employees may choose to leave the firm rather than exert more effort in a highly levered firm. However, the aggregate evidence leads us to conclude that debt provides an incentive effect for all employees, and that this effect is stronger when employment opportunities are worse. D. Imputed firm value, financial leverage and outside employment opportunities We next use Imputed Firm Value, based on the median value-to-productivity multiplier for the industry, as the dependent variable in the regression. We base multipliers on (i) Output per Employee and (ii) Output per Employee Hour. Table 9 presents results from the analysis of the relation between Leverage and Imputed Firm Value. The results are similar to those when Output per Employee or Output per Employee Hour is the dependent variable. The coefficient on

23 Leverage is positive and statistically significant at the 1% level for all estimations. The coefficients on the interaction terms with Quit Rate and Hire Rate are both negative and statistically significant at the 1% level, and the coefficient on the interaction term with Unemployment Growth Rate is significantly positive at the 1% level. To assess the economic significance and magnitude of these effects, we compute the change in productivity when leverage increases from the 25th percentile to the 75th percentile at two values of outside employment opportunities, the 25th and the 75th percentiles. According to our hypothesis and the results presented earlier, the change in productivity from an increase in Leverage should be lower when the outside employment opportunities are better (75th percentile). We estimate economic significance for the employment opportunity measures and present the results in Table 10. Panel A presents estimates based on Output per Employee and Panel B presents results based on Output per Employee Hour. As benchmarks for comparison, Panel C presents the findings from a similar analysis for selected control variables. In each panel, the first group of two columns presents the change in employee productivity, the second group of two columns presents the change in Imputed Firm Value ($ millions), and the third group of two columns reports the % changes in Imputed Firm Value. In the computations, we hold all the other independent variables constant at their mean values. In Panel A of Table 10, when the Quit Rate is at the 25th percentile value, increasing Leverage from the 25th to the 75th percentile increases Output per Employee by 10.41%; the analogous value for the Hire Rate is 11.1% and for the unemployment growth rate is 3.7%. When the value of Quit Rate or Hire Rate is at the 75th percentile (implying better employment opportunities), the corresponding increase in Output per Employee for a change in leverage from the 25th to the 75th percentiles is lower by more than half. When the value of the Unemployment

24 Growth Rate is at the 75th percentile (implying worse employment opportunities), an increase in leverage from the 25th to the 75th percentiles improves Output per Employee by 5.1%, which is 39% higher than the corresponding improvement at the 25th percentile. For Imputed Firm Value, when the employment opportunity measures are at the 25th percentile value, the change in firm value is in excess of $14 million for Quit Rate or Hire Rate and $5 million for unemployment growth rate. When the employment opportunities are at the 75th percentile, increasing Leverage improves firm value by about $7 million. The magnitudes of the % change in Imputed Firm Value are similar to those for Output per Employee. The results based on Output per Employee Hour, presented in Panel B, are similar. Comparing these magnitudes to those in Panel C shows that only Asset Intensity has a bigger impact on employee productivity in the case of Quit Rate or Hire Rate, and only Asset Intensity and Employees have a bigger impact on employee productivity in the case of unemployment growth rate. The influence on productivity of changes in Firm Age, Operating Leverage, and Herfindahl Index is less than the influence of debt. We also compute the respective economic significance of the productivity-leverage relation when outside employment opportunities are set to the 10th, 25th, 50th, 75th, and 90th percentiles, respectively. Figure 1.A presents the plot of these five economic significance values for Quit Rate, Figure 1.B presents the graphs for Hire Rate, and Figure 1.c presents the graphs for Unemployment Growth Rate. All three figures highlight the fact that the impact of Leverage on productivity and firm value is decreasing (increasing) when outside employment opportunities increase (decrease). The pattern is similar for measures based on Output per Employee and Output per Employee Hour. Furthermore, the reduction in the impact of Leverage on productivity appears to be economically significant. For example, an increase in the Quit Rate from the 10th percentile to the 90th percentile reduces the effect of Leverage on Output per

25 Employee (Imputed Firm Value) from about 11% ($16m) to less than 3% ($4m), a reduction of 73% (75%). In summary, our analysis suggests that the positive impact of Leverage on employee productivity and firm value is economically significant, and that the impact of Leverage is significantly reduced if employees have better outside employment opportunities. V.

Robustness Checks A. Industry-level Regressions As a control for endogeneity that might arise at the firm level, we estimate our tests at the industry level using the industry median values based on 2-digit SIC codes for each variable. The results from this analysis presented in Table 11 indicate that all the previous findings hold at the aggregate industry level. The positive Leverage-Productivity relation is weaker (stronger) when employees have more (fewer) outside opportunities; coefficients on the interaction terms are all statistically significant at levels ranging from 1% to 10%. B. Alternative model specifications B.1 Quadratic-form regression In prior sections, we present the results of tests that exclude firms in the top leverage decile to eliminate financially distressed firms. We now include firms in the top leverage decile and fit a non-linear relation between employee productivity and leverage by including Leverage2 as an additional independent variable. We hypothesize a (positive) concave relation: a positive relation at lower debt levels and a negative one at sufficiently high levels of debt. The intuition is that at high debt levels, the effects of financial distress will be large enough to cause lower marginal productivity gains as debt increases. We do not report the results from the benchmark

26 specification but note that the coefficient on Leverage is significantly positive and that on Leverage2 is significantly negative. This positive concave relation is consistent with our expectation that the disciplinary benefit of debt is offset by expected financial distress costs at sufficiently high debt levels. To explore the impact of outside employment opportunities on a concave LeverageProductivity relation, we augment equation 2 by including an interaction term between proxies for outside employment opportunities and Leverage2 in addition to the interaction term for Leverage. We present the results from these tests for Output per Employee in the first three columns of Table 12, and for Output per Employee Hour in the second three columns. For Quit Rate and Hire Rate, the coefficient on the interaction term between Leverage and the outside employment measure is significantly negative, and the coefficient on the interaction term with Leverage2 is significantly positive. The positive coefficient on the interaction term for Leverage2 is consistent with the notion that outside employment opportunities reduce the expected cost of financial distress. For Unemployment Growth Rate, the respective coefficients have the expected opposite signs. These findings indicate that increases (decreases) in outside employment opportunities dampen (strengthen) the concave productivity-leverage relation. Our main finding that more outside employment opportunities of employees weaken the disciplining role of debt remains robust in this alternative specification when the leverage ratio is lower than the inflection point (around 0.48). B.2 Additional Control Variables We also control for factors such as unionization, wage levels, external monitoring mechanisms, the work environment and non-pecuniary compensation, and employee pension plans that are also likely to influence productivity. We measure the degree of external monitoring

27 by the percentage of block holdings (5% or more of outstanding shares) of institutional investors obtained from the CDA Spectrum database of SEC 13-f filings which is available for years 1980 to 2007. We proxy for the expected wage by the average weekly earnings in the firm’s industry computed using the Current Employment Statistics (CES) survey from BLS. Pension plans, particularly defined benefit pension plans, provide employees with incentives to work harder and improve productivity (Ippolito, 1998) and, therefore, we include a dummy variable that equals one if the firm has a defined benefit plan. The information on defined benefit plans is available in Compustat (data item 287 or 296) from 1986 to 2007. To control for the effects of unionization on employee productivity, we include the variable Union Coverage, the fraction of workers in an industry who are covered by a collective bargaining agreement, as a control variable in our analysis.16 As a measure of good work environment and non-pecuniary compensation, we include a dummy variable, Best Company, which equals one if the firm is in Fortune Magazine’s list of “100 best companies to work for” (available from 1998 to 2007). We do not report these results in tables, but note that the coefficients on the interaction terms between Leverage and all three outside employment opportunities measures have the expected signs and are statistically significant at the 1% level. Thus, our results are robust to the inclusion of the additional control variables. C. Other Robustness Tests We present additional robustness tests for Output per Employee and Output per Employee hour, respectively, in Tables 13 and 14. To present the relevant information in the most space-efficient manner, we present only the coefficients on the test variables.

16

Baldwin (1983) argues that firms keep inefficient plants to discourage unions from bargaining for higher wages. Doucouliagos and Laroche (2003) find a positive relation between unionization and productivity for U.S. firms. Chen, Kacperczyk, and Ortiz-Molina (2010, 2011) argue that unions reduce the agency costs of financing.

28 C.1 Employment Opportunities Orthogonal to Leverage Possibly, employment opportunities and leverage co-vary with unobserved variables. To control for this possibility, we substitute the residual from a regression of outside employment opportunities on Leverage for the outside employment variable. By construction, the residual will be orthogonal to Leverage. The results, presented in Panel A of Tables 13 and 14, are consistent with our previously reported results. We find a positive relation between productivity and leverage, and this relation becomes stronger as employment opportunities decline. C.2 Median Regressions To address the possibility that skewness, particularly in Unemployment Growth Rate, could influence results, we estimate median regressions and present the results in Panel B of Table 13 and 14. The signs and significance levels on the coefficients are qualitatively similar to those previously reported. C.3 A Value-added Measure of Productivity Panel C in Tables 13 (14) presents results for the productivity measure, EBITDA per Employee (Employee Hour), which is the ratio of operating income before depreciation and amortization (data item 13) to the number of employees (employee hours). The results for this value-added measure are similar to those for our other productivity measures. C.4 Inclusion of Zero-debt Firms The analysis thus far includes only firms with positive debt levels because research indicates that zero-debt firms are systematically different from firms with debt. We relax this restriction and include zero-debt firms in the sample, and present our findings in Panel D of Table 13 and 14. Consistent with the previous findings, the coefficients on the interaction terms

29 are negative and statistically significant for both Quit Rate and Hire Rate, and positive and statistically significant for the Unemployment Growth Rate. C.5 Excluding firms in the top quartile In our main tests, we exclude firms in the top decile for Leverage. To ensure that our results are not sensitive to the decile cutoff, we exclude firms in the top quartile and repeat our analysis on the bottom 75% of the smaple by leverage. We present results for this alternative sample in Panel E of Table 13 and 14. Again, we find that the coefficients on the interaction terms are negative and statistically significant for Quit Rate and Hire Rate; and that on the interaction term for Unemployment Growth Rate is significantly positive. C.6 Unrestricted Compustat Sample Panel F in Tables 13 and 14 presents results for the entire non-utility and non-finance Compustat sample without imposing any additional restrictions on the data. The results are similar to those reported for the base sample. C.7 Quit and Hire Rates from 1970-1981 In the tests thus far, we use the Quit Rate and Hire Rate from the JOLTS data, which starts in 2000. The BLS also provides quit and hire rates from 1970 to 1981, but covers only the manufacturing industry. Further, these rates and are not directly comparable with the JOLTS data. We use the BLS data from the earlier period to estimate our tests from 1970-1981 and present the results in Panel G of Table 13 and 14. The signs on coefficients are similar to those reported for our main tests, but the significance levels are lower.

30 C.8 Excluding Firms with the Most and the Fewest Number of Employees The number of employees for firms in our sample ranges from three employees to 116,192 employees. We exclude firms with the number of employees in the bottom (less than 60 employees) and top (more than 15,991 employees) deciles and estimate our tests on the middle 80% of the sample by employees. Panel H of Table 13 and 14 presents the results. The coefficients and statistical significance on the interaction terms are similar to results reported in earlier sections, which suggests that our results are not driven by very small or large firms. C.9 An alternative control for M&A events In the analysis thus far, we omit firms with sales growth greater than 200% to control for the impact of M&A activities. Firms report the impact of M&A on their sales in their income statements (Compustat data item 249). We repeat all our reported tests on the subsample that excludes firm-years in which firms reported M&A sales impact and present the results in the Panel I of Table 13 and 14. The coefficient on the interaction term is negative and statistically significant at the 1% level for the Quit Rate and Hire Rate; and that for the Unemployment Growth Rate is positive and statistically significant at the 1% level. VI.

Concluding Remarks We propose that better outside employment opportunities weaken the efficacy of debt as a disciplining mechanism and worse outside employment opportunities strengthen the efficacy. As a direct test of the debt disciplining hypothesis, we analyze the relation between employee productivity and financial leverage and examine its interaction with changes in labor market conditions in a large sample of publicly held firms over a span of 31 years. In support of the premise that debt serves as a costly disciplining mechanism to mitigate agency conflicts, we find

31 that financial leverage exerts a positive influence on employee productivity. Consistent with our hypothesis, we find that the influence of financial leverage on employee productivity is weaker when employees have more outside employment opportunities and stronger when outside employment opportunities are worse. The results from our tests imply that employees compare the costs that they incur to lessen the likelihood of financial distress, for instance additional effort, to the transaction costs of leaving the firm. As outside employment opportunities increase, the relative costs of leaving the firm become less than the costs of additional effort, and debt becomes less effective as a disciplining device. Thus, our results for the influence of outside employment opportunities on the relation between productivity and financial leverage emphasize the fact that disciplining costs are borne by the agent. Collectively, the positive relation between employee productivity and financial leverage strongly support the debt disciplining arguments of Grossman and Hart (1982) and Jensen (1986). More broadly, our study suggests that researchers studying control and alignment mechanisms and policymakers and activists who seek to improve governance in publicly held firms should consider the interaction of the labor markets with a firm’s attempts to mitigate agency problems. Governance and other control mechanisms minimize agency conflicts because they impose costs on agents for behavior that is inconsistent with shareholder wealth maximization and reward behavior that is aligned with the objective of maximizing shareholder wealth. Our results for the influence of outside employment opportunities on the efficacy of debt as a disciplining mechanism highlight the importance of controlling for labor market conditions in studies that examine the usefulness of other governance and incentive alignment mechanisms. Our results also suggest that the constrained optimal equilibrium level of any governance

32 mechanism will vary across industries and over time with conditions in the labor markets. Hence, our findings highlight the influence of both financial and nonfinancial markets on the efficacy of these observed equilibrium outcomes.

33 References: Agrawal, A. K., and D. A. Matsa. “Labor Unemployment Risk and Corporate Financing Decisions.” New York University and Northwestern University Working Paper (2011). Baldwin, C. “Productivity and Labor Unions: An Application of the Theory of Self-enforcing Contracts.” Journal of Business, 56 (1983), 155 -185. Berk, J., R. Stanton, and J. Zechner. “Human Capital, Bankruptcy, and Capital Structure.” Journal of Finance, 65 (3) (2010), 891-926. Brander, J., and T. Lewis. “Oligopoly and Financial Structure: The Limited Liability Effect.” American Economic Review, 76 (1986), 956-970. Brealey, R. S. Myers, and F. Allen. Principles of Corporate Finance, (2006), McGraw-Hill Irwin, New York. Bronars, S., and D. Deere. “The Threat of Unionization, the Use of Debt, and the Preservation of Shareholder Wealth.” Quarterly Journal of Economics, 106 (1991), 231-53. Brynjolfsson, E., and L. Hitt. “Computing Productivity: Firm-level Evidence.” Review of Economics and Statistics, 85 (4) (2003), 793-808. Campello, M. “Debt Financing: Does It Boost or Hurt Performance in Product Markets?” Journal of Financial Economics, 82 (2006), 135-172. Cappelli, P., and K. Chauvin. “An Interplant Test of the Efficiency Wage Hypothesis.” Quarterly Journal of Economics, 106 (1991), 769-787. Chen, H., M. Kacperczyk, and H. Ortiz-Molina. “Do Non-financial Stakeholders Affect the Pricing of Risky Debt? Evidence from Unionized Workers.” Review of Finance, Forthcoming (2010). Chen, H., M. Kacperczyk, and H. Ortiz-Molina. “Labor Unions, Operating Flexibility, and the Cost of Equity?” Journal of Financial and Quantitative Analysis, 46 (2011), 25-58. Chevalier, J.. “Capital Structure and Product-market Competition: Empirical Evidence from the Supermarket Industry.” American Economic Review, 85 (3) (1995), 415-435. Dasgupta, S. and K. Sengupta. “Sunk Investment, Bargaining and Choice of Capital Structure.” International Economic Review, 34 (1993), 203–220. Denis, D., and D. Denis. “Managerial Discretion, Organizational Structure, and Corporate Performance.” Journal of Accounting and Economics, 16 (1993), 209-236. Doucouliagos, C. and P. Laroche. “What Do Unions Do to Productivity? A Meta-analysis.” Industrial Relations, 42 (2003), 650-691. Gardner, J. and C. Trzcinka. “All-equity Firms and the Balancing Theory of Capital Structure.” Journal of Financial Research, 15 (1992), 77-90. Gilson, S. “Management Turnover and Financial Distress.” Journal of Financial Economics, 25 (1989), 241-62. Gilson, S. “Bankruptcy, Boards, Banks, and Blockholders: Evidence on Changes in Corporate Ownership and Control when Firms Default.” Journal of Financial Economics, 27 (1990), 355-387. Gilson, S., and M. Vetsuypens. “CEO Compensation in Financially Distressed Firms: An Empirical Analysis.” Journal of Finance, 48 (1993), 425-58. Grossman, S. and O. Hart. “Corporate Financial Structures and Managerial Incentives.” in The Economics of Information and Uncertainty, (1982), ed. J. McCall, University of Chicago Press, Chicago. Hanka, G. “Debt and the Term of Employment.” Journal of Financial Economics, 48 (1998), 245-282. Hirsch, B., and D. Macpherson. “Union membership and coverage database from the current population survey: Note.” Industrial and Labor Relations Review, 56 (2003), 349-54. Hoberg, G., and G. Phillips. “Real and Financial Industry Booms and Busts.” Journal of Finance, 65 (1) (2010), 4586. Hottenrott, V., and S. Blank. “Assessing NAFTA-part II: The impact of NAFTA on jobs, wages, and income inequality.” Pace University working paper, (1998). Ippolito, R. Pension Plans and Employee Performance: Evidence, Analysis and Policy. Chicago: University of Chicago Press, (1998).

34 Jensen, M. “Agency Costs of Free Cash Flow, Corporate Finance, and Takeovers.” American Economic Review, 76 (1986), 323-329. Kale, J., and H. Shahrur. “Capital Structure and Characteristics of Supplier and Customer Markets.” Journal of Financial Economics, 83 (2007), 321-365. Kaplan, S. “The Effects on Management Buyouts on Operating Performance and Value.” Journal of Financial Economics, 24 (1989), 217-254. Kovenock, D., and G. Phillips. “Capital Structure and Product Market Rivalry: How Do We Reconcile the Theory and the Evidence.” American Economic Review, 85 (1995), 403-408. Kovenock, D., and G. Phillips. “Capital Structure and Product Market Behavior: An Examination of Plant Closing and Investment Decisions.” Review of Financial Studies, 10 (3) (1997), 767-803. Kurke, L., and H. Aldrich. “Mintzberg was Right! A Replication and Extension of the Nature of Managerial Work.” Management Science, (1983), 975-984. Leary, M., and M. Roberts. “Do Firms Rebalance Their Capital Structures?” The Journal of Finance, 60 (2005), 2575-2619. Lehn, K., and A. Poulsen. “Free Cash Flow and Stockholder Gains in Going Private Transactions.” Journal of Finance, 44 (1989), 771-788 Lehn, K., J. Netter, and A. Poulsen. “Dual-class Recapitalizations Versus Leveraged Buyouts.” Journal of Financial Economics, 27 (1990), 557-580. MacKay, P., and G. Phillips. “How Does Industry Affect Firm Financial Structure?” Review of Financial Studies, 18 (2005), 1433-1466. Maksimovic, V. “Capital Structure in Repeated Oligopolies.” RAND Journal of Economics, 19 (1988), 389-406. Maksimovic, V. “Product Market Imperfections and Loan Commitments.” Journal of Finance, 45 (5) (1990), 16411655. Marais, L., K. Schipper, and A. Smith. “Wealth Effects of Going Private for Senior Securities.” Journal of Financial Economics, 23 (1989), 155-191. Matsa, D. A. “Capital Structure as a Strategic Variable: Evidence from Collective Bargaining.” Journal of Finance, 65 (3) (2010), 1197-1232. McLaren, J., and S. Hakobyan. “Looking for local labor-market effects of the NAFTA.” NBER Working Paper, no. 16535, (2010). Muscarella, C., and M. Vetsuypens. “Efficiency and Organizational Structure: A Study of Reverse LBOs.” Journal of Finance, 45 (1990), 1389-1413. Myers, S. “The Determinants of Corporate Borrowing.” Journal of Financial Economics, 5 (1977), 147-175. Opler, T. “Operating Performance in Leveraged Buyouts: Evidence from 1985-1989.” Financial Management, 21 (1992), 37-41. Opler, T., and S. Titman. “The Determinants of Leveraged Buyout Activity: Free Cash Flow versus Financial Distress Costs.” Journal of Finance, 48 (1993), 1985-1999. Oyer, P. “Why Do Firms Use Incentives that Have No Incentive Effects?” Journal of Finance, 59 (2004), 16191649. Parrino, R. “CEO Turnover and Outside Succession: A Cross-sectional Analysis.” Journal of Financial Economics, 46 (1997), 165-197. Perotti, E., and C. Spier. “Capital Structure as a Bargaining Tool: The Role of Leverage in Contract Renegotiation.” American Economic Review, 83 (1993), 1131–1141. Phillips, G. “Increased Debt and Industry Product Markets: An Empirical Analysis.” Journal of Financial Economics, 37 (1995), 189-238. Rajgopal, S., T. Shevlin., and V. Zamora. “CEOs’ Outside Employment Opportunities and the Lack of Relative Performance Evaluation in Compensation Contracts.” Journal of Finance, 61 (2006), 1813-1844. Schoar, A. “Effects of Corporate Diversification on Productivity.” Journal of Finance, 57 (2002), 2379-2403. Schott, P. “U.S. Manufacturing Exports and Imports by SIC or NAICS Category and Partner Country, 1972 to 2005.” Yale University Working Paper, (2010).

35 Scott, R. “NAFTA’s Hidden Costs: Trade Agreement Results in Job Losses, Growing Inequality, and Wage Suppression for the United States, NAFTA at Seven: Its Impact on Workers in all Three Nations.” Economic Policy Institute Briefing Paper, (2001). Scott, R. “The High Price of “Free Trade”: NAFTA’s Failure Has Cost the United States Jobs across the Nation.” Economic Policy Institute Briefing Paper, (2003). Sharpe, S. “Financial Market Imperfections, Firm Leverage, and the Cyclicality of Employment.” American Economic Review, 84 (1994), 1060-1074. Smith, A. “Capital Ownership Structure and Performance: The Case of Management Buyouts.” Journal of Financial Economics, 27 (1990), 143-165. Strebulaev, I., and B. Yang. “The Mystery of Zero-leverage Firms.” Stanford University Working Paper, (2006). Stulz, R. “Managerial Discretion and Optimal Financing Policy.” Journal of Financial Economics, 26 (1990), 3-27. Titman, S. “The Effect of Capital Structure on a Firm’s Liquidation Decision.” Journal of Financial Economics, 13 (1984), 137-151.

FIGURE 1 The Impact of Outside Employment Opportunities on the Productivity-leverage Relation Figure 1. A; Quit Rate and changes in employee productivity when Leverage changes from 25th to 75th percentile. This figure presents changes in Output per Employee, Output per Employee Hour and Imputed Firm Values when Leverage changes from 25th to 75th percentile holding quit rate at different levels. Quit Rate is the number of employees in an industry that voluntarily quit their jobs divided by the total number of employees in the industry. We measure Output per Employee (Output per Employee Hour) as sales plus changes in inventories divided by the number of employees (hours). Imputed Firm Value – Employee (Hour) equals Output per Employee (Output per Employee Hour) multiplied by the industry median of firm value per unit of employee productivity. (% or $m)

Quit Rate (continued on next page)

FIGURE 1 (continued) The Impact of Outside Employment Opportunities on the Productivity-leverage Relation Figure 1. B: Hire rate and changes in employee productivity when Leverage changes from 25th to 75th percentile. This figure presents changes in Output per Employee, Output per Employee Hour and Imputed Firm Values when Leverage changes from 25th to 75th percentile holding hire rate at different levels. Hire rate is the total number of new hires in an industry divided by the total number of employees. We measure Output per Employee (Output per Employee Hour) as sales plus changes in inventories divided by the number of employees (hours). Imputed Firm Value – Employee (Hour) equals Output per Employee (Output per Employee Hour) multiplied by the industry median of firm value per unit of employee productivity.

(% or $m)

Hire Rate (continued on next page)

FIGURE 1 (continued) The Impact of Outside Employment Opportunities on the Productivity-leverage Relation Figure 1. C: Unemployment Growth Rate and changes in employee productivity when Leverage changes from 25th to 75th percentile. This figure presents changes in Output per Employee, Output per Employee Hour and Imputed Firm Values when Leverage changes from 25th to 75th percentile holding unemployment growth rate at different levels. Unemployment Growth Rate is the growth rate of the industry unemployment rate. We measure Output per Employee (Output per Employee Hour) as sales plus changes in inventories divided by the number of employees (hours). Imputed Firm Value – Employee (Hour) equals Output per Employee (Output per Employee Hour) multiplied by the industry median of firm value per unit of employee productivity. (% or $m)

Unemployment Growth Rate

TABLE 1 Descriptive Statistics The sample is from the Compustat database for 1976-2007 and excludes firms with no debt or debt ratios above the 90th percentile (0.671). Leverage is the book value of debt divided by the sum of the book value of debt and market value of common equity. Output per Employee (Employee Hour) is sales plus changes in inventories divided by the number of employees (total number hours the firm’s employees work in a year). Firm Value Multiplier – Employee (Hour) is the industry median firm value (market value of equity plus book value of debt) per unit of Output per Employee (Output per Employee Hour). Imputed Firm Value – Employee (Hour) equals Output per Employee (Hour) times the Firm Value Multiplier – Employee (Hour). EBITDA per Employee is the ratio of operating income before depreciation and amortization to the number of employees. Quit Rate is the number of employees in an industry that voluntarily quit their jobs divided by the total number of employees in the industry. Hire rate is the number of new hires in an industry divided by the total number of employees. Quit Rate and Hire Rate are available for 2000-2007. Unemployment Growth Rate is the growth rate of the industry unemployment rate. Employees is the number of employees reported in Compustat. Asset Intensity is total assets divided by the number of employees. Firm Age is the number of years the firm appears in Compustat. Operating Leverage is gross Property, Plant & Equipment divided by total assets. The Herfindahl Index is from (Hoberg and Phillips (2010) and covers both public and private firms. We winsorize all firm-level variables at the 1% and 99% values. Panel A. Summary Statistics Variables Obs. Mean Median Maximum Minimum Std. Dev. Leverage 103,394 0.238 0.200 0.671 0.000 0.191 Employee Productivity Output per Employee ($k) 103,394 239.154 163.711 1759.976 7.261 269.864 Firm Value Multiplier (000s) - Employee 103,394 1.221 0.770 8.968 0.123 1.388 Imputed Firm Value ($m) - Employee 103,394 320.668 138.053 4526.592 5.397 593.094 Output per Employee Hour ($) 81,705 127.234 85.709 961.850 2.797 146.425 Firm Value Multiplier ($m) – Hour 81,705 2.948 1.830 19.944 0.230 3.365 Imputed Firm Value ($m ) - Hour 81,705 369.482 157.389 4992.118 6.221 693.748 EBITDA per Employee ($k) 103,394 23.125 15.067 452.767 -289.329 85.269 Outside Employment Opportunity Quit Rate (%) 25,250 1.816 1.508 5.267 0.400 0.748 Hire Rate (%) 25,250 3.527 2.967 7.967 1.300 1.402 Unemployment Growth Rate (%) 98,917 1.511 -3.448 130.769 -55.000 29.363 Control Variables Employees (k) 103,394 6.915 1.000 116.192 0.003 17.455 Asset Intensity ($k) 103,394 300.612 141.982 3658.808 14.934 546.033 Firm Age 103,394 24.352 23 57 4 11.590 Operating Leverage 103,394 0.567 0.491 2.039 0.020 0.375 Herfindahl Index 103,394 0.064 0.054 0.916 0.019 0.042 (continued on next page)

TABLE 1 (continued) Descriptive Statistics Panel B. Pearson Correlations among Leverage, Employee Productivity and Outside Employment Opportunities Leverage

Output per Employee

Output per Imputed Firm Employee Hour Value–Emp.

Output per Employee

0.031***

Output per Employee Hour

0.045***

0.932***

Imputed Firm Value–Emp.

0.055***

0. 530***

0.537***

Imputed Firm Value–Hour

0.075***

0.487***

0.496***

0.901***

EBITDA per Employee

0.057***

0.560***

0.552***

0.315 ***

Imputed Firm Value–Hour

EBITDA per Employee

Quit Rate

0.296***

Quit Rate

-0.039**

-0.090***

-0.056***

-0.014***

Hire Rate

-0.037***

-0.043***

-0.022***

-0.040***

-0.031***

0.014**

0.911***

0.032***

-0.010***

-0.007*

-0.016**

-0.008**

-0.038***

-0.071***

Unemployment Growth Rate

Hire Rate

0.002

-0.034***

-0.071***

TABLE 2 Median Employee Productivity, Leverage, and Outside Employment Opportunities by Industry This table presents the median employee productivity by industry. The sample of firms is from the Compustat database for 1976-2007 and excludes firms with no debt or debt ratios above the 90th percentile (0.671). Output per Employee is the ratio of sales plus changes in inventories to the number of employees. Imputed Firm Value equals Output per Employee multiplied by the industry median of firm value per unit of employee productivity. We define firm value as market value of equity plus book value of debt. Quit Rate is the number of employees in an industry that voluntarily quit their jobs divided by the total number of employees in the industry. Hire rate is the total number of new hires in an industry divided by the total number of employees. Quit Rate and Hire Rate are available for 2000-2007. Unemployment Growth Rate is the growth rate of the unemployment rate by industry. The industry descriptions for 2-digit SIC codes are from the U.S. Census Bureau. We present industries in descending order of Output per Employee. Output per Output per Quit Hire Unemployment Employee Employee Leverage Rate Rate Growth Rate SIC Industry Description ($k) Hour ($) (%) (%) (%) 29 Petroleum and coal products 820.70 399.49 0.29 1.58 2.80 -7.32 15 General building contractors 545.54 239.32 0.43 2.33 6.47 -7.88 51 Wholesale trade--nondurable goods 410.86 209.11 0.29 1.50 2.83 -3.85 50 Wholesale trade--durable goods 350.70 189.83 0.30 1.48 2.83 -3.85 13 Oil and gas extraction 349.86 178.49 0.23 1.60 3.83 -3.23 12 Coal mining 319.83 142.81 0.36 1.60 3.83 -1.89 21 Tobacco manufactures 244.67 168.00 0.22 1.52 2.80 4.60 14 Nonmetallic minerals, except fuels 236.94 105.27 0.22 1.60 3.83 -3.23 33 Primary metal industries 227.14 114.99 0.33 1.20 2.40 -8.89 47 Transportation services 226.22 113.45 0.13 1.52 3.08 -6.25 20 Food and kindred products 225.47 111.47 0.24 1.58 2.80 0.00 44 Water transportation 223.56 112.11 0.37 1.52 3.08 -4.44 26 Paper and allied products 211.03 102.56 0.32 1.58 2.80 -8.89 48 Communications 204.60 111.07 0.29 1.52 3.08 -6.67 52 Building materials, hardware, garden 204.00 108.66 0.26 3.10 4.84 -3.51 supply & mobile 28 Chemicals and allied products 201.95 99.95 0.11 1.58 2.81 0.00 54 Food stores 186.82 108.36 0.36 2.98 4.88 -3.03 10 Metal mining 186.80 88.65 0.13 1.60 3.83 -4.55 57 Furniture, home furnishings and 186.11 103.90 0.27 3.10 4.88 -3.17 equipment stores 78 Motion pictures 176.10 114.07 0.28 2.64 7.29 -3.70 16 Heavy construction contractors 174.18 90.51 0.20 2.31 6.32 -7.11 32 Stone, clay, glass, and concrete products 172.53 90.35 0.33 1.20 2.40 -2.00 55 Automotive dealers and gasoline service 172.42 87.82 0.36 2.98 4.88 -3.51 stations 45 Transportation by air 172.03 84.98 0.38 1.50 3.08 -6.06 24 Lumber and wood products 168.09 78.07 0.28 1.58 2.81 -8.06 35 Industrial machinery and equipment 167.91 85.55 0.18 1.20 2.40 0.00 30 Rubber and miscellaneous plastics 166.71 87.52 0.25 1.20 2.40 1.33 products (continued on next page)

TABLE 2 (continued) Median Employee Productivity, Outside Employment Opportunities and Leverage by Industry

SIC Industry Description 27 59 37 17 34 39 73 38 36 25 42 53 22 23 31 87 56 76 82 75 80 79 72 70 58 83

Printing and publishing Miscellaneous retail Transportation equipment Special trade contractors Fabricated metal products Miscellaneous manufacturing industries Business services Instruments and related products Electrical and electronic equipment Furniture and fixtures Motor freight transportation and warehousing General merchandise stores Textile mill products Apparel and other textile products Leather and leather products Engineering and management services Apparel and accessory stores Miscellaneous repair services Educational services Automotive repair, services, and parking Health services Amusement and recreational services Personal services Hotels, rooming houses, camps, and other lodging Eating and drinking places Social services

Output per Output per Quit Employee Employee Leverage Rate Hour ($k) (%) ($)

Hire Industry Rate Unemployment (%) Growth %

162.14 159.81 156.60 153.04 152.68 148.48 147.53 146.47 141.31 124.43 124.29

84.65 96.22 78.59 79.16 76.61 85.72 77.67 76.59 75.14 62.57 59.17

0.18 0.21 0.27 0.22 0.27 0.24 0.08 0.12 0.15 0.23 0.30

1.58 3.10 1.20 2.29 1.20 1.20 2.70 1.20 1.20 1.58 1.48

2.80 4.88 2.40 6.32 2.40 2.40 5.61 2.40 2.40 2.80 3.08

-4.65 -3.51 -3.13 -5.68 -5.56 -7.58 -2.08 -3.85 -3.45 -8.51 -6.25

121.43 121.32 121.19 120.83 108.99 105.43 105.14 103.86 103.56 92.45 78.60 69.26 55.43

83.02 63.93 73.77 62.75 56.26 72.97 54.66 51.64 47.09 48.01 45.29 37.89 33.30

0.34 0.38 0.25 0.22 0.12 0.14 0.23 0.10 0.45 0.23 0.32 0.23 0.41

3.10 1.58 1.58 1.20 2.73 2.98 1.70 1.16 4.80 1.88 2.64 2.70 4.80

4.88 2.80 2.81 2.40 5.65 4.88 2.20 2.57 6.94 3.06 7.16 5.61 6.94

-3.51 -5.49 -4.44 -5.15 0.00 -3.51 -1.75 -1.69 -2.27 -2.63 -3.70 -2.08 -3.70

40.18 28.49

26.12 16.57

0.25 0.37

3.10 1.83

4.88 3.01

-3.37 -2.63

TABLE 3 Outside Employment Opportunities, Employee Productivity, and Leverage: Industry Median Adjusted Analysis This table presents OLS regression results of the employee productivity-leverage relation and the effect of outside employment opportunities on the productivityleverage relation. Except for Leverage, we adjust all continuous firm-level variables for their respective industry medians in each year to control for industry effects. The sample is from the Compustat database for 1976-2007 and excludes firms with no debt or debt ratios above the 90th percentile (0.671). The dependent variable is the natural log of Output per Employee or Output per Employee Hour, as designated in the top row. Table 1 presents definitions for all variables. We present standard errors, adjusted for heteroskedasticity and firm clustering, in parentheses. *** significant at 1% level; ** significant at 5% level; * significant at 10% level. Output per Employee Hour Output per Employee

Leverage

Base 0.143*** (0.017)

Unemployment Growth Rate 0.144*** (0.018)

Quit Rate 0.505*** (0.081)

Hire Rate 0.578*** (0.086)

Base 0.169*** (0.021)

Unemployment Growth Rate 0.167*** (0.021)

Quit Rate 0.543*** (0.086)

Hire Rate 0.597*** (0.093)

Leverage × Outside Emp. Opportunities

0.176*** (0.032)

-0.139*** (0.039)

-0.092*** (0.022)

0.228*** (0.042)

-0.144*** (0.042)

-0.091*** (0.024)

Outside Employment Opportunities

-0.061*** (0.012)

0.028** (0.014)

0.017** (0.008)

-0.083*** (0.014)

0.042*** (0.015)

0.026*** (0.008)

Ln(Employees)

0.021*** (0.003)

0.022*** (0.003)

0.025*** (0.004)

0.025*** (0.004)

0.028*** (0.003)

0.029*** (0.003)

0.033*** (0.005)

0.033*** (0.005)

Ln(Asset Intensity)

0.540*** (0.007)

0.539*** (0.007)

0.522*** (0.012)

0.522*** (0.012)

0.537*** (0.008)

0.538*** (0.008)

0.525*** (0.012)

0.525*** (0.012)

Ln(Firm Age)

0.055*** (0.008)

0.057*** (0.008)

0.084*** (0.014)

0.084*** (0.014)

0.057*** (0.009)

0.058*** (0.009)

0.095*** (0.014)

0.094*** (0.014)

Operating Leverage

0.053*** (0.015)

0.060*** (0.015)

0.171*** (0.025)

0.172*** (0.025)

0.069*** (0.017)

0.073*** (0.017)

0.180*** (0.027)

0.180*** (0.027)

Herfindahl Index

0.087 (0.092)

0.089 (0.093)

0.430*** (0.156)

0.423*** (0.156)

0.225** (0.111)

0.228** (0.111)

0.707*** (0.188)

0.720*** (0.190)

Intercept

-0.171*** (0.013)

-0.145*** (0.013)

-0.266*** (0.035)

-0.275*** (0.038)

-0.080*** (0.018)

-0.031* (0.018)

-0.201*** (0.037)

-0.218*** (0.040)

Year Fixed Effects Industry-median-adjusted Firm-level Variables Observations R2

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

103,394 0.340

98,917 0.338

25,250 0.304

25,250 0.304

81,705 0.311

79,969 0.311

24,620 0.284

24,620 0.284

TABLE 4 Outside Employment Opportunities, Employee Productivity and Leverage: Industry Fixed Effects Analysis This table presents the OLS regression results of employee productivity-leverage relation and the effect of outside employment opportunities on the productivityleverage relation. The sample is from the Compustat database for 1976-2007 and excludes firms with no debt or debt ratios above the 90th percentile (0.671). The dependent variable is the natural log of Output per Employee or Output per Employee Hour, as designated in the top row. Table 1 presents definitions for all variables. We present standard errors, adjusted for heteroskedasticity and firm clustering, in parentheses. *** significant at 1% level; ** significant at 5% level; * significant at 10% level. Output per Employee

Leverage

Base 0.143*** (0.017)

Unemployment Growth Rate 0.144*** (0.018)

Output per Employee Hour

Quit Rate Hire Rate 0.505*** 0.578*** (0.081) (0.086)

Base 0.131*** (0.021)

Unemployment Growth Rate 0.129*** (0.022)

Quit Rate Hire Rate 0.401*** 0.506*** (0.092) (0.099)

Leverage × Outside Emp. Opportunities

0.176*** (0.032)

-0.139*** -0.092*** (0.039) (0.022)

0.232*** (0.042)

-0.085* (0.047)

-0.075*** (0.027)

Outside Employment Opportunities

-0.061*** (0.012)

0.028** (0.014)

-0.088*** (0.014)

0.060** (0.024)

0.089*** (0.018)

0.017** (0.008)

Ln(Employees)

0.021*** (0.003)

0.022*** (0.003)

0.025*** 0.025*** (0.004) (0.004)

0.035*** (0.003)

0.035*** (0.003)

0.036*** (0.005)

0.036*** (0.005)

Ln(Asset Intensity)

0.540*** (0.007)

0.539*** (0.007)

0.522*** 0.522*** (0.012) (0.012)

0.534*** (0.008)

0.535*** (0.008)

0.520*** (0.012)

0.521*** (0.012)

Ln(Firm Age)

0.055*** (0.008)

0.057*** (0.008)

0.084*** 0.084*** (0.014) (0.014)

0.065*** (0.010)

0.066*** (0.010)

0.097*** (0.015)

0.096*** (0.015)

Operating Leverage

0.053*** (0.015)

0.060*** (0.015)

0.171*** 0.172*** (0.025) (0.025)

0.072*** (0.016)

0.075*** (0.016)

0.179*** (0.026)

0.179*** (0.026)

Herfindahl Index

0.087 (0.092)

0.089 (0.093)

0.430*** 0.423*** (0.156) (0.156)

0.222* (0.130)

0.228* (0.130)

0.494** (0.211)

0.529** (0.211)

Intercept

-0.171*** (0.013)

-0.145*** (0.013)

-0.266*** -0.275*** (0.035) (0.038)

1.067*** (0.227)

1.055*** (0.236)

0.814*** (0.119)

0.681*** (0.121)

Year Fixed Effects Industry Fixed Effects Observations R2

Yes Yes 103,394 0.340

Yes Yes 98,917 0.338

Yes Yes 25,250 0.304

Yes Yes 25,250 0.304

Yes Yes 81,705 0.490

Yes Yes 79,969 0.489

Yes Yes 24,620 0.461

Yes Yes 24,620 0.461

TABLE 5 NAFTA and the Productivity-Leverage Relation This table presents the OLS results of the effect of NAFTA on the productivity-leverage relation. The sample is from the Compustat database for 1976-2007 and excludes firms with no debt or debt ratios above the 90th percentile (0.671). Except for Leverage, we adjust all continuous firm-level variables for the respective industry median in each year. The dependent variable is the natural log of Output per Employee (Hour), which we define as sales plus changes in inventories divided by the number of employees (hours). Leverage is the ratio of book value of debt divided by the sum of book value of debt plus market value of common equity. NAFTA equals one for observations in year 1994 and onwards, and equals zero otherwise. Leverage*NAFTA is the product of Leverage and the NAFTA dummy. Manufacturing firms include all firms with SIC codes between 2000 and 3999. Non-manufacturing firms include all firms with SIC codes smaller than 2000, and larger than 3999. Table 1 presents definitions for all other variables. We present standard errors, adjusted for heteroskedasticity and firm clustering, in parentheses. *** significant at 1% level; ** significant at 5% level; * significant at 10% level. Output per Employee Leverage Leverage × NAFTA NAFTA Ln(Employees) Ln(Asset Intensity) Ln(Firm Age) Operating Leverage Herfindahl Index Intercept Industry-median-adjusted Firm-level variables Observations R2

Output per Employee Hour

All Firms Mfg. Firms Non-Mfg. Firms 0.135*** 0.161*** 0.089*** (0.020) (0.024) (0.033) 0.130*** 0.236*** 0.022 (0.030) (0.041) (0.045) -0.002 -0.034** 0.003 (0.010) (0.014) (0.017) 0.027*** 0.036*** 0.018*** (0.003) (0.003) (0.004) 0.537*** 0.508*** 0.553*** (0.007) (0.011) (0.009) 0.042*** 0.050*** 0.025*** (0.007) (0.010) (0.010) 0.060*** 0.002 0.094*** (0.015) (0.021) (0.020) 0.138 0.729*** -0.072 (0.090) (0.168) (0.107) -0.162*** -0.219*** -0.109*** (0.011) (0.016) (0.017)

All Firms Mfg. Firms Non-Mfg. Firms 0.138*** 0.137*** 0.132*** (0.027) (0.036) (0.039) 0.170*** 0.286*** 0.024 (0.035) (0.047) (0.051) -0.022* -0.050*** 0.010 (0.012) (0.016) (0.018) 0.033*** 0.041*** 0.028*** (0.003) (0.004) (0.004) 0.534*** 0.500*** 0.553*** (0.008) (0.012) (0.010) 0.063*** 0.072*** 0.041*** (0.008) (0.012) (0.011) 0.070*** -0.013 0.127*** (0.017) (0.025) (0.023) 0.245** 0.910*** 0.016 (0.109) (0.201) (0.129) -0.068*** -0.125*** -0.016 (0.014) (0.020) (0.020)

Yes

Yes

Yes

Yes

Yes

Yes

103,394 0.326

58,954 0.276

44,440 0.378

81,705 0.301

43,995 0.254

37,710 0.348

TABLE 6 NAFTA and the Productivity-Leverage Relation: Subsamples based on the Changes in Mexico Tariff Rates Before and After NAFTA This table presents the OLS results of the effect of NAFTA on the productivity-leverage relation. The sample includes manufacturing firms from the Compustat database for 1976-2007 and excludes firms with no debt or debt ratios above the 90th percentile (0.671). Manufacturing firms are firms with SIC codes between 2000 and 3999. Except for Leverage, we adjust all continuous firm-level variables for the respective industry median in each year. Tariff Changes are computed as an industry’s Mexico tariff rate in 2000 minus the industry’s Mexico tariff rate in 1990. The dependent variable is the natural log of Output per Employee or Output per Employee Hour, which we define as sales plus changes in inventories divided by the number of employees or employee hours. Leverage is the ratio of book value of debt divided by the sum of book value of debt plus market value of common equity. NAFTA equals one for observations in year 1994 and onwards, and equals zero otherwise. Table 1 presents definitions for all other variables. We present standard errors, adjusted for heteroskedasticity and firm clustering, in parentheses. *** significant at 1% level; ** significant at 5% level; * significant at 10% level. Output per Employee Output per Employee Hour Least or No Moderate Most Least or No Moderate Most Reduction Reduction Reduction Tariff Tariff Reduction Reduction Reduction Tariff Tariff in Tariff in Tariff in Tariff Change Change in Tariff in Tariff in Tariff Change Change (Top 30%) (Middle 40%) (Bottom 30%) Median 0.494*** 0.047 (0.107) (0.041) 0.770*** 0.109** (0.161) (0.053) -0.055*** 0.005 (0.020) (0.018) 0.066*** 0.025*** (0.004) (0.005) 0.475*** 0.579*** (0.014) (0.013) 0.046*** 0.061*** (0.014) (0.013) -0.041 -0.012 (0.030) (0.029) 1.415*** 0.041 (0.256) (0.203) -0.295*** -0.095*** (0.023) (0.021)

Ind. Union Coverage ≤Median 0.256*** (0.063) 0.264*** (0.082) -0.072*** (0.026) 0.067*** (0.007) 0.383*** (0.022) 0.036* (0.021) 0.018 (0.041) 1.107*** (0.366) -0.306*** (0.033)

Leverage >Median 0.134*** (0.039) 0.198*** (0.052) -0.026 (0.021) 0.004 (0.004) 0.538*** (0.013) 0.068*** (0.013) -0.024 (0.028) 0.243 (0.204) -0.057** (0.022)

Output per Employee Hour Ind. Union Leverage Coverage ≤Median >Median 0.590*** 0.018 (0.174) (0.043) 0.747*** 0.165*** (0.217) (0.055) -0.064** 0.005 (0.025) (0.019) 0.075*** 0.024*** (0.005) (0.005) 0.465*** 0.575*** (0.017) (0.014) 0.067*** 0.077*** (0.017) (0.014) -0.060* -0.044 (0.036) (0.030) 1.782*** 0.171 (0.314) (0.212) -0.218*** -0.001 (0.030) (0.022)

Ind. Union Coverage ≤Median 0.246*** (0.063) 0.311*** (0.084) -0.087*** (0.027) 0.069*** (0.007) 0.383*** (0.023) 0.056*** (0.022) -0.003 (0.043) 1.203*** (0.373) -0.211*** (0.034)

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

29,477 0.380

29,477 0.236

29,557 0.393

29,397 0.201

21,997 0.351

21,998 0.219

21,900 0.382

22,095 0.186

TABLE 9 Imputed Firm Value, Financial Leverage and Outside Employment Opportunities This table presents OLS results on the relation between the imputed firm value and financial leverage. The sample is from the Compustat database for 1976-2007 and excludes firms with no debt or debt ratios above the 90th percentile (0.671). Except for Leverage, we adjust all continuous firm-level variables for their respective industry medians in each year to control for industry effects. The dependent variable is the natural log of Imputed Firm Value – Employee (Hour), which equals Output per Employee (Hour) multiplied by the industry median of firm value per unit of employee productivity. Firm value is defined as market value of equity plus book value of debt. Table 1 presents definitions for all other variables. We present standard errors, adjusted for heteroskedasticity and firm clustering, in parentheses. *** significant at 1% level; ** significant at 5% level; * significant at 10% level.

Leverage

Imputed Firm Value based on Imputed Firm Value based on Output per Employee Output per Employee Hour Unemployment Unemployment Growth Rate Quit Rate Hire Rate Growth Rate Quit Rate Hire Rate 0.156*** 0.537*** 0.606*** 0.162*** 0.526*** 0.581*** (0.018) (0.083) (0.088) (0.021) (0.085) (0.091)

Leverage × Outside Employment Opportunities

0.180*** (0.033)

-0.148*** (0.040)

-0.096*** (0.023)

0.229*** -0.138*** (0.041) (0.041)

-0.087*** (0.024)

Outside Employment Opportunities

-0.063*** (0.012)

0.026* (0.014)

0.017** (0.008)

-0.082*** 0.037** (0.014) (0.014)

0.023*** (0.008)

Ln(Employees)

0.022*** (0.003)

0.025*** (0.005)

0.025*** (0.005)

0.030*** 0.034*** (0.003) (0.005)

0.034*** (0.005)

Ln(Asset Intensity)

0.549*** (0.007)

0.532*** (0.012)

0.532*** (0.012)

0.529*** 0.516*** (0.008) (0.012)

0.516*** (0.012)

Ln(Firm Age)

0.055*** (0.008)

0.086*** (0.014)

0.086*** (0.014)

0.056*** 0.094*** (0.009) (0.014)

0.093*** (0.014)

Operating Leverage

0.065*** (0.015)

0.178*** (0.026)

0.179*** (0.026)

0.072*** 0.175*** (0.017) (0.026)

0.176*** (0.026)

Herfindahl Index

0.070 (0.096)

0.419*** (0.161)

0.408** (0.161)

0.210* (0.109)

0.678*** (0.179)

0.686*** (0.180)

Intercept

-0.036*** (0.013)

-0.137*** (0.036)

-0.147*** (0.039)

-0.029 (0.018)

-0.191*** (0.037)

-0.206*** (0.040)

Year Fixed Effects Industry-median-adjusted Firmlevel Variables Observations R2

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

98,917 0.333

25,250 0.295

25,250 0.295

79,969 0.311

24,620 0.285

24,620 0.285

TABLE 10 Economic Significance This table presents changes in Output per Employee and Imputed Firm Value when Leverage, outside employment opportunities and other control variables change from the 25 th percentile to the 75th percentile based on the OLS estimates of models in Table 3 and Table 9. The sample is from the Compustat database for 1976-2007 and excludes firms with no debt or debt ratios above the 90th percentile (0.671). Quit Rate is the number of employees in an industry that voluntarily quit their jobs divided by the total number of employees in the industry. Hire rate is the total number of new hires in an industry divided by the total number of employees. Quit and Hire rates are only available from 2000 to 2007. We set all control variables to their mean values and vary only the variable of interest. Output per Employee (Output per Employee Hour) equals sales plus changes in inventories divided by the number of employees (hours). Imputed Firm Value – Employee (Hour) equals Output per Employee (Hour) multiplied by the industry median of firm value per unit of employee productivity. Firm value is the market value of equity plus book value of debt. Leverage is the ratio of book value of debt divided by the sum of book value of debt plus market value of common equity. Panel C presents the economic significance for control variables based on the estimates from the Quit Rate (Unemployment Growth Rate) regressions. Table 1 presents definitions for all other variables. Panel A. Changes in productivity when Leverage changes from 25th to 75th percentile Change in output per Change in imputed firm Change in imputed firm employee (%) value ($m) - Employee value (%) - Employee 25th 75th 25th 75th 25th 75th Outside options are held at: Percentile Percentile Percentile Percentile Percentile Percentile Quit Rate 10.41 4.93 14.81 7.13 11.10 5.23 Hire Rate 11.11 4.43 15.63 6.44 11.72 4.71 Unemployment Growth Rate 3.67 5.10 5.41 7.27 4.04 5.50 Panel B. Changes in productivity when Leverage changes from 25 th to 75th percentile Change in Output per Change in imputed firm Change in imputed firm Employee Hour (%) value ($m) - Hour value (%) - Hour 25th 75th 25th 75th 25th 75th Outside options are held at: Percentile Percentile Percentile Percentile Percentile Percentile Quit Rate 11.48 5.75 16.82 8.86 11.17 5.69 Hire Rate 11.85 5.19 17.46 8.23 11.65 5.28 Unemployment Growth Rate 4.16 6.02 6.27 9.04 4.00 5.86 Panel C. Changes in productivity when control variables change from 25 th to 75th percentile

Ln(Employees) Ln(Asset Intensity) Ln(Firm Age) Operating Leverage Herfindahl Index

Estimates based on Coefficients from Quit Rate Regressions Change in Employee Change in imputed Productivity (%) firm value ($m) Employee Hour Employee Hour 6.45 8.52 8.79 13.44 41.58 41.82 48.26 53.55 6.65 7.52 9.27 11.48 5.53 5.82 7.85 8.77 1.07 1.76 1.46 2.67

Change in imputed firm value (%) Employee Hour 6.45 8.78 42.38 41.11 6.80 7.44 5.76 5.66 1.04 1.69

Estimates based on Coefficients from Unemployment Growth Rate Regressions Ln(Employees) Ln(Asset Intensity) Ln(Firm Age) Operating Leverage Herfindahl Index

5.68 42.94 4.51 1.94 0.22

7.49 42.86 4.59 2.36 0.57

7.24 46.09 5.59 2.72 0.23

11.26 51.58 6.56 3.47 0.79

5.68 43.73 4.35 2.10 0.17

7.74 42.14 4.43 2.33 0.52

TABLE 11 Employ Productivity, Leverage, and Outside Employment Opportunities: Industry Level Regressions This table presents OLS results from the regressions on the industry level. The sample is from the Compustat database for 1976-2007 and excludes firms with no debt or debt ratios above the 90th percentile (0.671). All variables equal their respective industry medians (based on 2-digit SIC codes) in each year. The dependent variable is the natural log of Output per Employee (Output per Employee Hour), which is sales plus change in inventories divided by the number of employees (hours). Leverage is the ratio of book value of debt divided by the sum of book value of debt plus market value of common equity. Table 1 presents definitions for all other variables. We present standard errors, adjusted for heteroskedasticity and industry clustering, in parentheses. *** significant at 1% level; ** significant at 5% level; * significant at 10% level. Output per Employee Output per Employee Hour Unemployment Unemployment Growth Rate Quit Rate Hire Rate Growth Rate Quit Rate Hire Rate Leverage

0.361*** (0.087)

0.883*** (0.299)

0.786** (0.348)

0.229** (0.107)

0.976** (0.406)

0.790* (0.441)

Leverage × Outside Employment Opportunities

0.838** (0.396)

-0.304*** (0.108)

-0.137* (0.073)

0.792* (0.465)

-0.487*** (0.136)

-0.213** (0.088)

Outside Employment Opp.

-0.226* (0.122)

0.052 (0.039)

0.013 (0.021)

-0.276* (0.147)

0.164*** (0.062)

0.044 (0.032)

Ln(Employees)

0.099*** (0.011)

0.079*** (0.020)

0.076*** (0.020)

0.129*** (0.013)

0.072*** (0.026)

0.076*** (0.026)

Ln(Asset Intensity)

0.640*** (0.019)

0.662*** (0.030)

0.664*** (0.030)

0.539*** (0.024)

0.574*** (0.041)

0.560*** (0.041)

Ln(Firm Age)

0.161*** (0.028)

0.118* (0.065)

0.129* (0.066)

0.101*** (0.033)

0.100 (0.094)

0.095 (0.096)

Operating Leverage

-0.673*** (0.042)

-0.601*** (0.076)

-0.610*** (0.077)

-0.481*** (0.056)

-0.299*** (0.108)

-0.296*** (0.108)

Herfindahl Index

-0.806*** (0.133)

-0.156 (0.134)

-0.160 (0.134)

-0.590*** (0.136)

-0.059 (0.149)

-0.022 (0.148)

Intercept

1.868*** (0.114)

1.628*** (0.270)

1.651*** (0.264)

1.840*** (0.152)

1.088** (0.448)

1.339*** (0.434)

Year Fixed Effects

Yes

Yes

Yes

Yes

Yes

Yes

Observations

1,776

460

460

1,667

449

449

R2

0.666

0.711

0.709

0.516

0.523

0.517

TABLE 12 Employ Productivity, Leverage, and Outside Employment Opportunities: Quadratic Form This table presents the OLS results of the impact of outside employment opportunities on the productivity-leverage relation using the quadratic-form specification. We include Leverage2 and Leverage2×Outside Employment Opportunities in the model. The sample is from the Compustat database for 1976-2007 and excludes firms with no debt. The dependent variable is the natural log of Output per Employee (Hour), which we define as sales plus changes in inventories divided by the number of employees (hours). Table 1 presents definitions for all other variables. Except for predicted Leverage related variables, we adjust all continuous firm-level variables for the respective industry median in each year. We present standard errors, adjusted for heteroskedasticity and firm clustering, in parentheses. *** significant at 1% level; ** significant at 5% level; * significant at 10% level.

Leverage

Output per Employee Output per Employee Hour Unemployment Unemployment Growth Rate Quit Rate Hire Rate Growth Rate Quit Rate Hire Rate 0.422*** 0.995*** 1.125*** 0.464*** 1.014*** 1.120*** (0.040) (0.177) (0.192) (0.047) (0.188) (0.206)

Leverage2

-0.436*** (0.047)

-0.888*** (0.192)

-0.984*** (0.211)

-0.464*** (0.055)

-0.860*** -0.943*** (0.208) (0.232)

Leverage × Outside Employment Opportunities

0.357*** (0.078)

-0.276*** (0.084)

-0.179*** (0.048)

0.457*** (0.097)

-0.255*** -0.163*** (0.090) (0.053)

Leverage2 × Outside Employment Opportunities Outside Employment Opp.

-0.266*** (0.092)

0.267*** (0.090)

0.165*** (0.054)

-0.357*** (0.115)

0.230** (0.101)

-0.081*** (0.014)

0.037** (0.016)

0.023** (0.009)

-0.106*** (0.016)

0.049*** 0.031*** (0.017) (0.009)

Ln(Employees)

0.017*** (0.002)

0.021*** (0.004)

0.021*** (0.004)

0.024*** (0.003)

0.029*** 0.029*** (0.005) (0.005)

Ln(Asset Intensity)

0.541*** (0.007)

0.526*** (0.011)

0.526*** (0.011)

0.544*** (0.008)

0.533*** 0.533*** (0.012) (0.012)

Ln(Firm Age)

0.054*** (0.007)

0.075*** (0.013)

0.075*** (0.013)

0.055*** (0.009)

0.084*** 0.084*** (0.014) (0.014)

Operating Leverage

0.054*** (0.014)

0.168*** (0.024)

0.169*** (0.024)

0.072*** (0.016)

0.185*** 0.185*** (0.025) (0.025)

Herfindahl Index

0.049 (0.091)

0.361** (0.150)

0.354** (0.150)

0.166 (0.108)

0.596*** 0.610*** (0.181) (0.181)

Intercept

-0.059*** (0.015)

-0.173*** (0.037)

-0.185*** (0.039)

-0.078*** (0.019)

-0.220*** -0.240*** (0.040) (0.044)

Year Fixed Effects Industry-median-adjusted Firmlevel Variables Observations R2

Yes

Yes

0.143** (0.061)

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

108,997 0.350

27,724 0.316

27,724 0.316

87,776 0.324

27,022 0.298

27,022 0.298

TABLE 13 Output per Employee, Leverage, and Outside Employment Opportunities: Robustness Checks This table presents regression results of the impact of outside employment opportunities on the productivity-leverage relation for a set of robustness checks. The dependent variable is the natural log of Output per Employee, which is sales plus change in inventories divided by the number of employees. In A, we first regress outside employment opportunity proxies on leverage and use the residuals in place of outside employment opportunity proxies when examine the productivity-leverage relation. In B, we present results based on a median regression. In C, the dependent variable is EBITDA per Employee, which is the ratio of operating income before depreciation and amortization to the number of employees. In D, we include zero-debt firms and excludes firms with debt ratios above the 90th percentile (0.671). In E, we exclude zero-debt firms and firms with debt ratios above the 75th percentile (0.459). In F, the sample includes all firm years in Compustat without imposing any additional restrictions. In G, we use quit rates and hire rates from the BLS job survey done in 1970 to 1981. In H, the sample excludes firms in the top (> 15,991) and bottom (< 60) deciles by number of employees. In I, we excludes firms that report M&A impact on sales. Table 1 presents definitions for all other variables. Except for Leverage related variables, we adjust all continuous firm-level variables for the respective industry median in each year. All regressions include control variables and year fixed effects as in previous tables. We present standard errors, adjusted for heteroskedasticity and firm clustering, in parentheses. *** significant at 1% level; ** significant at 5% level; * significant at 10% level. Unemployment Unemployment Unemployment Growth Rate Quit Rate Hire Rate Growth Rate Quit Rate Hire Rate Growth Rate Quit Rate Hire Rate A. Residual Regression

B. Median Regression

C. EBITDA per Employee

Leverage

0.146*** 0.256*** 0.256*** (0.018) (0.035) (0.035)

0.022*** 0.140*** 0.155*** (0.007) (0.035) (0.043)

24.722*** 73.655*** 65.975*** (7.336)) (23.181) (22.453)

Leverage × Outside Employment Opportunities

0.191*** -0.144*** -0.095*** (0.033) (0.039) (0.022)

0.120*** -0.062*** -0.037*** (0.024) (0.018) (0.011)

21.085* (11.978)

Outside Employment Opportunities

-0.064*** 0.029** 0.018** (0.012) (0.014) (0.008)

-0.038*** -0.009* (0.008) (0.005)

-0.004 (0.003)

-10.325*** -2.419 (3.164) (3.288)

Observations

98,917

25,250

25,250

98,917

25,250

25,250

98,917

25,250

25,250

R2 /Pseudo R2

0.338

0.304

0.304

0.230

0.211

0.211

0.160

0.151

0.151

D. Includes Zero Debt Firms

E. Bottom 75% by Leverage

F.

-19.986** (8.660)

-7.939* (4.526) 1.265 (1.778)

Unrestricted Compustat Sample

Leverage

0.161*** 0.526*** 0.584*** (0.018) (0.077) (0.082)

0.238*** 0.547*** 0.610*** (0.025) (0.120) (0.000)

0.134*** 0.333*** 0.387*** (0.014) (0.054) (0.059)

Leverage × Outside Employment Opportunities

0.195*** -0.141*** -0.089*** (0.030) (0.037) (0.021)

0.241*** -0.129** -0.085** (0.050) (0.059) (0.033)

0.122*** -0.051** (0.022) (0.025)

-0.041*** (0.015)

Outside Employment Opportunities

-0.070*** 0.030** 0.017** (0.010) (0.012) (0.007)

-0.068*** 0.028* 0.017** (0.013) (0.015) (0.008)

-0.049*** 0.022* (0.009) (0.012)

0.013* (0.007)

Observations 2

R

11,2944

31,061

31,061

83,149

22,201

22,201

134,212

38,748

38,748

0.320

0.291

0.292

0.326

0.293

0.293

0.328

0.296

0.296

Table 13 (continued) Output per Employee, Leverage, and Outside Employment Opportunities: Robustness Checks Quit Rate

Hire Rate

G. Quit/Hire Rates from 1970-1981

Unemployment Growth Rate Quit Rate Hire Rate H. Middle 80% by Employees

Unemployment Growth Rate Quit Rate I.

Hire Rate

Alternative Control for M&A Events

Leverage

0.115** (0.051)

0.107* (0.059)

0.083*** 0.453*** 0.519*** (0.018) (0.083) (0.087)

0.197*** 0.570*** (0.019) (0.087)

0.653*** (0.093)

Leverage × Outside Employment Opportunities

-0.050** (0.026)

-0.031 (0.021)

0.224*** -0.139*** -0.091*** (0.033) (0.041) (0.022)

0.143*** -0.145*** (0.035) (0.042)

-0.098*** (0.024)

Outside Employment Opportunities

0.023** (0.011)

0.014 (0.008)

-0.068*** 0.037*** 0.020** (0.012) (0.014) (0.008)

-0.042*** 0.030** (0.013) (0.015)

0.020** (0.009)

Observations

20,278

20,278

79,781

19,867

19,867

90,156

23,089

23,089

0.424

0.424

0.385

0.338

0.339

0.319

0.289

0.289

2

R

TABLE 14 Output per Employee Hour, Leverage, and Outside Employment Opportunities: Robustness Checks This table presents the regression results of the impact of outside employment opportunities on the productivity-leverage relation for a set of robustness checks. The dependent variable is the natural log of Output per Employee Hour, which is sales plus change in inventories divided by the number of employee hours. In A, we first regress outside employment opportunity proxies on leverage and use the residuals in place of outside employment opportunity proxies when examine the productivity-leverage relation. In B, we present results based on a median regression. In C, the dependent variable is EBITDA per Employee Hour, which is the ratio of operating income before depreciation and amortization to the number of employee hours. In D, we include zero-debt firms and excludes firms with debt ratios above the 90th percentile (0.671). In E, we exclude zero-debt firms and firms with debt ratios above the 75th percentile (0.459). In F, the sample includes all firm years in Compustat without imposing any additional restrictions. In G, we use quit rates and hire rates from the BLS job survey done in 1970 to 1981. In H, the sample excludes firms in the top (> 15,991) and bottom (< 60) deciles by number of employees. In I, we excludes firms that report M&A impact on sales. Table 1 presents definitions for all other variables. Except for Leverage related variables, we adjust all continuous firm-level variables for the respective industry median in each year. All regressions include control variables and year fixed effects as in previous tables. We present standard errors, adjusted for heteroskedasticity and firm clustering, in parentheses. *** significant at 1% level; ** significant at 5% level; * significant at 10% level. Unemployment Unemployment Unemployment Growth Rate Quit Rate Hire Rate Growth Rate Quit Rate Hire Rate Growth Rate Quit Rate Hire Rate A. Residual Regression

B. Median Regression

C. EBITDA per Employee

Leverage

0.170*** 0.282*** 0.279*** (0.021) (0.037) (0.037)

0.021** 0.169*** 0.151** (0.008) (0.049) (0.055)

14.762*** (1.178))

35.801*** 33.201*** (5.314) (5.567)

Leverage × Outside Employment Opportunities

0.245*** -0.151*** -0.094*** (0.042) (0.042) (0.024)

0.176*** -0.055** -0.024 (0.028) (0.025) (0.015)

19.068*** (2.754)

-4.433* (2.321)

-1.533 (1.258)

Outside Employment Opportunities

-0.086*** 0.044*** 0.027** (0.014) (0.015) (0.008)

-0.054*** -0.011* (0.009) (0.006)

-0.006* (0.004)

-5.550*** (0.933)

0.679 (0.876)

0.704 (0.516)

Observations

79,969

24,620

24,620

79,969

24,620

24,620

79,969

24,620

24,620

R2 /Pseudo R2

0.311

0.284

0.284

0.212

0.197

0.197

0.137

0.147

0.147

D. Includes Zero Debt Firms

E. Bottom 75% by Leverage

F.

Unrestricted Compustat Sample

Leverage

0.195*** (0.021)

0.536*** 0.575*** (0.081) (0.088)

0.273*** 0.576*** 0.626*** (0.030) (0.127) (0.139)

0.173*** (0.017)

0.358*** 0.395*** (0.058) (0.065)

Leverage × Outside Employment Opportunities

0.239*** (0.038)

-0.129*** -0.078*** (0.040) (0.023)

0.319*** -0.121* (0.063) (0.063)

0.137*** (0.027)

-0.045 (0.028)

-0.034** (0.017)

Outside Employment Opportunities

-0.086*** (0.012)

0.038** (0.013)

0.022)** (0.007)

-0.093*** 0.041** 0.026*** (0.016) (0.016) (0.009)

-0.060*** (0.011)

0.029** (0.013)

0.018** (0.007)

-0.077** (0.036)

Observations

92,608

30,351

30,351

67,936

21,669

21,669

110,661

37,840

37,840

R2

0.292

0.272

0.272

0.300

0.275

0.275

0.306

0.283

0.283

Table 14 (continued) Output per Employee Hour, Leverage, and Outside Employment Opportunities: Robustness Checks Quit Rate

Hire Rate

G. Quit/Hire Rates from 1970-1981

Unemployment Growth Rate Quit Rate Hire Rate H. Middle 80% by Employees

Unemployment Growth Rate Quit Rate I.

Hire Rate

Alternative Control for M&A Events

Leverage

0.139** (0.068)

0.147* (0.075)

0.068*** 0.419*** 0.442*** (0.021) (0.086) (0.092)

0.228*** 0.611*** (0.023) (0.092)

0.678*** (0.100)

Leverage × Outside Employment Opportunities

-0.027 (0.026)

-0.021 (0.021)

0.289*** -0.121*** -0.069*** (0.041) (0.043) (0.024)

0.191*** -0.149*** (0.045) (0.045)

-0.097*** (0.026)

Outside Employment Opportunities

0.024** (0.010)

0.017** (0.008)

-0.085*** 0.033** (0.014) (0.014)

0.020** (0.008)

-0.062*** 0.045*** (0.015) (0.016)

0.029** (0.009)

Observations

4,830

4,830

63,077

19,199

19,199

72,741

22,514

22,514

0.460

0.459

0.372

0.339

0.339

0.292

0.270

0.270

2

R